Detailed Design Report

Transcript

Detailed Design Intelligent Vehicle Initiative Specialty Vehicle Field Operational Test August 2001 Mn/DOT – US DOT Cooperative Agreement Task 3.5 of Cooperative Agreement DTFH61-99-X-00101 TABLE OF CONTENTS 1.0 Project Overview......................................................................................................................................1 2.0 Infrastructure Design...............................................................................................................................7 3.0 Vehicle Design ........................................................................................................................................37 4.0 Driver Assistive System (DAS) Interface Design.................................................................................85 5.0 vehDAQ Design ....................................................................................................................................139 Appendix A: Lane Awareness System Appendix B: Pavement Surface Preparation & Application Techniques Appendix C: Communication Interface Standard Appendix D: Installation Instructions to Lane Awareness System Appendix E: Altra Technologies Side Radar Communication Specifications Appendix F: Routes and Randomization for Human Factors Experiments 1 & 2 Appendix G: Human Factors Instructions & Questionaires for Experiments 1 & 2 Appendix H: ANOVA Summary Tables for Experiment 2 Appendix I: Main Instructions (Rosemount Field Study) Detailed Design Report ii August 10, 2001 TABLE OF FIGURES, PICTURES, AND TABLES Figure 1. Figure 2. Figure 3. Figure 4. Figure 5. Figure 6. Figure 7. Figure 8. Figure 9. Figure 10. Figure 11. Figure 12. Figure 13. Figure 14. Figure 15. Figure 16. Figure 17. Figure 18. Figure 19. Figure 20. Figure 21. Figure 22. Figure 23. Figure 24. Figure 25. Figure 26. Figure 27. Figure 28. Figure 29. Figure 30. Figure 31. Figure 32. Figure 33. Figure 34. Figure 35. Figure 36. Figure 37. Figure 38. Driver Assistive System Block Diagram..................................................................2 Map of the test corridor State Highway 7...............................................................5 Magnetic lateral system components.......................................................................7 State Highway 7 Corridor between Hutchinson and Silver Lake ........................9 McLeod County 7 roadway ......................................................................................10 Overall visibility system block diagram. .................................................................16 Highway 7 visibility and weather sensor configuration.........................................17 Weather station foundation design..........................................................................19 MRVI computation data flow diagram...................................................................20 Point contrast to distance relation ...........................................................................24 Regions of interest in the image ...............................................................................24 GPS correction base station configuration .............................................................27 Corridor showing locations of GPS base stations ..................................................29 Timing of GPS base stations along Highway 7.......................................................29 RF Signal propagation for GPS base stations ........................................................30 Vehicle System Block Diagram ................................................................................37 GPS Vehicle block diagram......................................................................................38 Heading estimate from successive GPS position solutions ....................................39 Simplified block diagram of the GPS/IMU integrated system..............................49 Graphical representation of a geospatial database used for radar processing ...54 Solid model results of the snowplow radar configuration study...........................55 Data flow for vehicle PC 104 computers .................................................................64 Simplified HUD system layout .................................................................................65 GPS mount for snowplows .......................................................................................71 Mounting system for magnetic sensor for typical snowplows...............................72 Eaton Vorad Forward –looking radar mounts ......................................................74 Top structure is easily modified...............................................................................76 Design drawing of the combiner mount........................................................................77 Plan layout typical of the equipment configuration for a snowplow....................79 The U.S. Standard National Network Signs..................................................................88 Overall distribution of lane departure durations ..................................................100 Overall distribution of logarithmically transformed lane departure durations .101 Histogram showing mean speeds, Test 1.................................................................111 Histogram showing mean speeds, Test 2.................................................................112 Histogram showing mean speeds, Test 3.................................................................113 Histogram showing mean speeds, Test 4.................................................................114 Histogram showing mean speeds, Test 5.................................................................115 Increase in the mean speed at which the participants drove ................................116 Detailed Design Report iii August 10, 2001 Figure 39. Figure 40. Figure 41. Figure 42. Figure 43. Figure 44. Average lane departure durations, singular modality...........................................118 Average lane departure durations, dual modality .................................................119 Block diagram of the data acquisition system ........................................................126 Course and the six checkpoints for traffic control .................................................128 vehDAQ system functional block diagram .............................................................140 Functional block diagram of the quad unit ............................................................142 Picture 1: Picture 2: Picture 3: Picture 4: Picture 5: Picture 6: Picture 7: Picture 8: Picture 9: Picture 10: Picture 11: Picture 12: Picture 13: Picture 14: Picture 15: Picture 16: Picture 17: Picture 18: Picture 19: Picture 20: Picture 21: Picture 22: Picture 23: Picture 24: Picture 25: Picture 26: Picture 27: Picture 28: Picture 29: Picture 30: Picture 31: Picture 32: Picture 33: Picture 34: Magnetic Tapes..........................................................................................................8 Standard Two Lane...................................................................................................12 Two Lane with Center Turns...................................................................................12 Three Lane to Two Lane ..........................................................................................13 Four Lane with Center Median and Right Turn ...................................................13 Chanhassen Melody Hill Water Tower...................................................................32 Mayer Water Tower..................................................................................................33 Silver Lake Water Tower .........................................................................................34 Head Up Display Image captured at 30 MPH on Highway 101 ...........................41 Crossbow HDX IMU.................................................................................................43 Leica SR530 GPS Receiver.......................................................................................43 Leica SR530 GPS receiver, Front View ..................................................................44 Leica AT 502 Dual Frequency GPS antenna..........................................................44 Antenex Phantom RF antenna.................................................................................45 GLB SNRDS RF Modem..........................................................................................46 Magnetic sensor bar and connecting cable .............................................................50 Hardware Interface...................................................................................................52 Box Top ......................................................................................................................56 Front end of radar power and communication interface box...............................56 Back end of radar power and communication interface box ................................57 Altra Technology’s Proximity Detector for side object detection.........................57 Front panel of Vehicle PC104 based vehicle computer .........................................61 Back of PC 104 computer .........................................................................................62 Inside of PC 104 vehicle computer...........................................................................63 Potato Head figure illustrating properties of the combiner ..................................66 HUD display under bright conditions .....................................................................67 LED based side collision warning interface............................................................69 Audible User Interface Hardware ...........................................................................70 Short I beam magnetic sensor mount......................................................................73 Combiner in folded up position ...............................................................................78 Combiner in normal operating position..................................................................78 Snowplow equipment cabinet...................................................................................80 Same box with protective covering ..........................................................................80 220 Amp Leece-Neville Alternator ..........................................................................81 Detailed Design Report iv August 10, 2001 Picture 35: A female XLR connector ..........................................................................................82 Picture 36: Power bus faceplate .....................................................................................................82 Picture 37: ExelTech 1100 Watt Inverter ...................................................................................83 Picture 38: No-warning control condition ......................................................................................90 Picture 39: Red line warning condition........................................................................................91 Picture 40: Double line warning condition .....................................................................................91 Picture 41: White area warning condition..................................................................................92 Picture 42: Red area warning condition .....................................................................................92 Picture 43: Driving with the DAS in the driving simulator ......................................................95 Picture 44: Cameras used for capturing the driver’s feet, hands, and face, respectively ......141 Picture 45: Forward looking, high resolution camera...............................................................141 Picture 46: Front of vehDAQ computer .....................................................................................144 Picture 47: Removable hard driver carrier................................................................................145 Picture 48: Business end of vehDAQ computer .........................................................................145 Table 1. Table 2. Table 3. Table 4. Table 5. Table 6. Table 7. Table 8. Tabel 9. Table 10. Tabel 11. Tabel 12. Tabel 13. Tabel 14. Tabel 15. Location of Weather Stations along Highway 7 .....................................................18 Base station equipment list.......................................................................................28 GPS Vehicle Parts list ...............................................................................................41 PC 104 Vehicle computer components ....................................................................60 Audible User Interface hardware ............................................................................69 Summary of 2x2 ANOVA .........................................................................................102 Result of post hoc test showing the effect of lane departure warnings.................102 Single-modality—summary of ANOVAs ................................................................117 Dual-modality—summary of ANOVAs ..................................................................117 Percentage of Lane–crossing responses...................................................................120 Average response to questions about the DAS Interface.......................................121 Average response to questions about the lane departure warnings .....................122 Average response to warnings via dual modalities/combination ..........................122 List of peripheral vehDAQ components .......................................................................143 vehDAQ computer components ...............................................................................144 Detailed Design Report v August 10, 2001 1.0 PROJECT OVERVIEW The system under evaluation for the US DOT Specialty Vehicle Generation Zero Field Operational Test is designed to provide a driver a means to maintain desired lane position and avoid collisions with obstacles during periods of low visibility. This program is motivated by the fact that specialty vehicles often must operate under inclement weather conditions. Typically associated with these inclement weather conditions are low visibility situations. The driver assistive system improves safety for the specialty vehicle operator by providing the necessary cues for lane keeping and collision avoidance normally unavailable during poor visibility conditions. The driver assistive system, when placed in public safety vehicles, also improves safety conditions for the general public by facilitating all-weather emergency services, and in the case of snowplows, opening roads and keeping them passable in heavy weather for other emergency vehicles and the general motoring public. The primary thrust of the project will be snowplow vehicles; however, ambulances and police vehicles will also be included. The project will implement, operate and evaluate all necessary infrastructure, in-vehicle sensing technology, in-vehicle processing including algorithms, and driver-vehicle interfaces. Testing of these systems will take place on state and county highways using state and county vehicles under low- visibility conditions such as snow, blowing snow, fog, and night. Human factors laboratory testing will be done to assure the driver-vehicle interface systems are based on the best possible human-centered design. The project results will be used to provide the Federal Highway Administration and an independent evaluator with data and to inform decision makers and the general public of the potential for these systems to improve the safety and productivity of the transportation system. Vehicle positioning, collision avoidance, and the driver interface constitute the primary components of the driver assistive system. Vehicle positioning is accomplished through a combination of a Differential GPS (DGPS) – geo-spatial database system and a roadway magnetic tape/sensor based system. Collision warning and avoidance is accomplished with radar sensors and signal processing techniques which take advantage of information returned by the vehicle positioning system. Finally, information is provided to the driver via the driver interface system, which will employ graphical, haptic, and auditory interfaces (or any combination thereof as deemed appropriate by the human factors work) to provide an optimal information path to the driver. A block diagram of the driver assistive system illustrating components and signal paths is shown in Figure 1. The system works as follows: DGPS and the magnetic tape system provide information regarding the position of the vehicle; DGPS provides global information, and magnetic tape system provides local information in Detailed Design Report 1 August 10, 2001 Inertial Measurement Unit acceleration, rotation rates Driver Visual Interface position Magnetic Sensor lane lateral position Geospatial Database Processor Navigation Processor local landscape orientation Satellite Signals GPS Corrections GPS receiver Driver Interface Processor vehicle global position Radar Processor forward sensors relevant obstacles side sensors rear sensors Figure 1. Driver assistive system block diagram Detailed Design Report Driver Haptic Interface 2 August 10, 2001 Driver Audible Interface the form of a lateral displacement of the sensor from the magnetic tape1. Vehicle orientation data (i.e., vehicle yaw, roll, and pitch rates, lateral, longitudinal, and vertical acceleration) is provide by an Inertial Measurement Unit (IMU). Given vehicle position and orientation, the geo-spatial database is queried to determine the presence and location of all relevant items in the local landscape. Simultaneously, the vehicle forward, side, and rear looking radar scans the environment local to the vehicle to detect the presence and location of obstacles around the vehicle. (In the forward view, presence and location means range, range rate, and azimuth angle to the sensed obstacles; in the side and rear view, range within the radar beam is all that is sensed.) The radar processor accepts this raw sensor data, and compares it to the results of the geo-spatial database query. The radar processor then determines radar returns which are a threat to a driver and which returns are associated with fixed elements of the infrastructure. When both the geo-spatial database processor and the radar processor complete their respective tasks, the results are sent to the driver interface processor. The driver interface processor determines whether the vehicle is in its proper lane, the probability of a collision with another vehicle or element of the local, and whether all sensors are functioning properly. If the driver is in no danger, no warnings are issued, and the driver is provided continuous assistance through the visual, haptic, and audible displays present in the vehicle.2 If the driver is heading for an undesired lane departure or collision with another object, the appropriate warning is issued in order that the driver take appropriate action. Each of the primary system components may be associated with sensors, infrastructure, processors, and displays. For instance, the vehicle positioning system infrastructure includes magnetic tape embedded in the roadway along the skip line and edge lines, a network of GPS receivers, antennae, and RF modems used to broadcast the GPS correction signals to the proper GPS receivers located on the test vehicles. The geospatial database, although resident on each vehicle, can also be considered infrastructure because it locates and provides attributes for each relevant item located near the roadway. These details will be further described in the specific system requirements in the sequel. A key component of the field operational test but not part of the driver assistive system is the data acquisition system. The purpose of the field operational test is to provide substantiated evidence that these systems do work as proposed, and that safety and operational benefits are achieved at a favorable cost:benefit ratio. Data acquisition capability will be present both in-vehicle and as part of the test corridor infrastructure. First, the driver assistive system has been proposed as a means with which to help a driver under conditions of low visibility. It is therefore imperative to document visibility 1 Lateral lane position is a subset of global position when the geo-spatial database is used. Used together, system robustness can be improved should one or the other system fail or become unavailable. Proprietary techniques are being developed to improve this system. 2 Precisely how these displays will be used and configured will be determined by the extensive human factors work being undertaken. Detailed Design Report 3 August 10, 2001 during the course of the field operational test. Visibility will be documented in two ways: the first means is with infrastructure-based weather stations; the second means is with an in-vehicle system utilizing a forward looking camera. Along the test corridor, a number of weather stations will be installed and networked to provide weather and visibility3 data to a central server. These weather stations will report atmospheric conditions including precipitation types, rates, and moisture content in addition to visibility measures. Local to each vehicle will be a vehicle data acquisition system (vehDAQ4). The heart of the vehDAQ system is the ability to accept video input from four cameras, multiplex the four video signals, compress the video signals, and then store the compressed captured images in real time to a hard disc mounted on each test vehicle. In addition to the video data, the vehDAQ also records audio data and engineering unit data (i.e., GPS position, magnetic tape lane position data, IMU data, radar “hits,” etc.) in synchronization with the video data. Three of the four video cameras are aimed at the driver or the driver interface; the fourth camera is aimed out of the windshield. The camera aimed out the windshield records the image as perceived by the driver. As will be explained in the visibility system subsection of this report, the images captured via the windshield camera will be used to determine a “motorist’s visibility index” which quantifies the effective visibility available to the driver at any time during the field operational test. The operational test will take place primarily on State Highway 7 between Hutchinson, MN to the west and I-494 in Minnetonka, MN to the east. Because of the involvement of McLeod County in the field operational test with their drivers and snowplow, sections of County Road 79 and County Road 7 near State Highway 7 will also be included in the field operational test. Magnetic tape will be installed on State Highway 7 between Hutchinson, MN and Silver Lake, MN and, tentatively (pending negotiations between the County and State) on a section of County Road 7 to be determined. The entirety of State Highway 7 between Hutchinson and the Twin Cities will be mapped for use with the GPS system, as will County Roads 7 and 79. A map of the State Highway 7 corridor is shown in Figure 2. 3 The visibility measure provided by the weather station is “meteorological visual range,” which indicates the density of the atmospheric particles that affect the visible-spectrum transfer function of air space between two separated points where the measurements take place. 4 This was termed “microDAS” in the TSR. However, a search turned up other products called “microDAS.” To avoid possible trademark problems and because the design described herein is new, the name vehDAQ has been assigned to the four camera based data acquisition unit. Detailed Design Report 4 August 10, 2001 Figure 2. Map of the test corridor: Trunk Highway 7 This Detailed Design report provides a detailed description of the major infrastructure, vehicle and related systems for this project. The report is organized as follows: First, infrastructure detailed designs are be presented. Infrastructure for this field operational test includes devices to measure visibility, a means to support GPS-based vehicle guidance through GPS corrections and a geo-spatial database, and a means to support magnetic tape-based vehicle guidance. Second, vehicle system detailed designs are provided. Vehicle systems for this field operational test include the vehicle GPS system, vehicle magnetic guidance system, vehicle radar system, vehicle data acquisition system, and the requirements for integrating these systems. Finally, the detailed designs for the driver forward interface, side interface, and rear interface are provided. These latter detailed designs precisely document how the information is provided to the driver via the three human interfaces. Detailed Design Report 5 August 10, 2001 Detailed Design Report 6 August 10, 2001 2.0 INFRASTRUCTURE DESIGN 2.1 MAGNETIC TAPE 2.1.1 System Overview The magnetic lateral guidance system consists of two main components: 1. Infrastructure Components: Magnetic pavement marking tape installed with warranty. 2. Vehicle Components: Magnetic sensor(s) with mounting hardware and output interface installed with warranty. Infrastructure Magnetic Pavement Marking Tape Vehicle Components Mounting Hardware Sensing Hardware Interface Hardware Figure 3. Magnetic Lateral Guidance System Components The system uses a continuous magnetic marking material with a permanent magnetic field pattern applied to the pavement surface. In snowplow areas the tape can be applied under the last lift of the asphalt surface or grooved into existing asphalt or concrete surfaces. The magnetic pavement marking tape has a thickness of about 1.3 [mm] and a width of 10 [cm]. The intrinsic coercively is greater that 4000 oersteds to insure that the tape is not inadvertently remagnetized by large magnets used to clear highways of metal debris. The field pattern alternates up and down, north and south, through the thickness of the tape at fixed distances along the tape length. A magnetic sensor is used to detect the tape to provide lateral location information of the vehicle on a road. The sensor unit contains multiple magnetoresistive elements oriented in orthogonal directions. Analog signals from the sensor are converted into digital values and processed by the computer module to obtain an estimated distance calculation. This information is then supplied to a driver interface representing the vehicle location with respect to the magnetic tape. The placement of the sensor on a snowplow has to account for the primary blade and its rotation, wing blades in various positions, and under body blades. Sensor placement must also maximize the sensor’s physical protection and minimize the magnetic interference from the plow blade. Detailed Design Report 7 August 10, 2001 Sensor placement on the ambulance and law enforcement vehicles would ideally be located at the far edges of the front bumper to obtain the maximum detection range outside the vehicle body. Currently there will be only one sensor installed on the ambulance and law enforcement vehicles located at the far edge of the front bumper on either the passenger or driver side. Placement on the passenger side would give the operator an indication of the road edge. Placement on the driver side would give the operator an indication of the centerline or skip line. The initial sensor location will be on the driver side. This placement will allow the operation of vehicles on Hwy. 7 and County Road 7 since both locations will have skip line tape installed. The final placement will be determined during the test stage of the program. For this project the magnetic tape will be grooved into the concrete highway surface along the skip and edge lines to create a magnetic boundary for each lane along the corridor. Grooving is necessary in snowplow areas to insure that the tape is not removed during snowplow operations. The center lane markings will be used for snowplow removal operations. Edge and skip lines will be used by the other vehicles for lane keeping. The tape configuration will be based on standard road marking procedures to include solid white and yellow tapes along with standard skip line configurations. Black tape will be used in areas that are not normally marked, such as access road intersections and portions of turn lanes. The “Preliminary Product Bulletin for Snow Removal Operations” is included as Appendix A. Picture 1 shows the currently available white, yellow and black tapes. White and yellow skip line tape is also available but not pictured. Picture 1: Magnetic tapes Detailed Design Report 8 August 10, 2001 Site Location and Material Requirements Magnetic tape will be installed along the edge and skip lines of the State Highway 7 corridor between Hutchinson and Silver Lake. This section, approximately 8 miles in length, has been reconstructed as a super-2 highway configuration. This will create a section of highway where the lanes are defined by the magnetic tape. Figure 4 is a map indicating the State Highway 7 corridor to be used. This portion of the project will require approximately 50 rolls of black magnetic tape 70 rolls of white skip line tape, 75 rolls of solid yellow tape, 125 rolls of yellow skip line tape and 400 rolls of solid white tape. Each roll of tape is 60 meters long. This is equivalent to approximately 27 lineal miles of tape. The installation will also require 70 pails of P50 adhesive and 20 pails of E44T adhesive. Each pail of adhesive contains 5 gallons of material. The E44T will be used at intersections to reduce the amount of tape damage experienced in high traffic turning areas. All other areas will use the P50 adhesive. Additional adhesive will be shipped if required. The northern and southern most lanes along the corridor will be marked. Black magnetic tape will be used to extend the edge line through right turn lanes and to extend the white skips through merging lanes. Black tape will also be used to supplement the white skips used for left turn lanes. Eastern End N Western End Figure 4. Trunk Highway 7 corridor between Hutchinson & Silver Lake Magnetic tape will also be installed along a County Road 7 corridor. The County Highway 7 corridor is shown in Figure 5. It runs N/NE from the intersection of County 79 to the intersection of 230th. This installation will require approximately 3.2 miles of solid yellow, 1.2 miles of yellow skip magnetic tape, and ten five-gallon pails (total of 50 gallons) of P50 pre-adhesive. The tape will replace the western most mid-road markings to form a continuous magnetic reference for the lateral guidance system. Detailed Design Report 9 August 10, 2001 Northern End N County 7 Southern End Figure 5. McLeod County 7 roadway 2.1.2. Installation Procedure Tape installation will be supervised by 3M Technical Service to insure that the proper installation procedures are used. The Draft Information Folder 5.19 “Pavement Surface Preparation and Application Techniques for 3M Magnetic Tape Series 2000 for Snow Removal Applications” is included as Appendix B. The critical specifications to note are the groove depth and groove location with regard to lane break line. The groove should be at a depth of 0.1 inches +10% / -0%. This insures that the tape will be at or below the road surface to prevent removal during snow removal operations. The groove location with respect to the lane break line should be nominally 2 inches from the break line. This prevents edge damage to the concrete surface. The tape installation will be monitored to insure proper adhesion, adequate magnetic signal strength and reflectivity in all temperature and road conditions. A tape log will be kept for the installation and will include tape installation location and environmental conditions, inspection dates, times, and environmental conditions, magnetic tape field Detailed Design Report 10 August 10, 2001 strengths for specified locations, tape adhesion comments if appropriate, reflectivity measurements and any additional notes. 2.1.3. Tape Installation Layout There are a number of different road configurations along the Highway 7 corridor that must be addressed. Each of these road configurations will require the use of different tape configuration combinations to achieve the desired magnetically outlined lanes. Five of these configurations are shown in pictures – see pictures 2-5 below. The appropriate tape configuration will be supplied by 3M to address all lane-marking configurations along the corridor. All markings will be per Minnesota Department of Transportation Construction Plan, Federal Project Number STPF 078-2. The Construction Plan will be used by 3M and the tape installer to determine and verify the marking requirements for this project. The plans will be returned to Mn/DOT upon completion of the installation. Again, all markings will be per state code requirements. The following discussions outlines modifications not covered by state code requirements. Picture 2 shows a standard two-lane road. Markings would normally include solid white magnetic tape for the edge lines and a yellow skip line for the center lane. There are cases, however, where the two lane road would require a combination of either two double solid yellow lines or one solid yellow line and one yellow skip line. In this case only one of the lines would be replaced with magnetic tape. The exact line to be replaced will be selected on site to insure a smooth, continuous magnetic reference line. Picture 3 shows a two-lane road with a center turn lane. In this case the existing lane marking would be extended through the turn lane using the black magnetic tape as indicated. There would normally not be any lane marking in this area. Picture 4 shows a three-lane merging into a two-lane. In this case the black tape is projected through the merge lane to achieve a continuous magnetic reference. The solid yellow tape continues until the lane is merged and then the appropriate marking is continued. This could be either a yellow skip, double solid yellow, or one solid yellow with one yellow skip. Picture 5 is a four-lane with center median and right turn lane. In this case the outer lanes are marked with solid white edge lines and white skip lines. A black tape is used to extend the edge line through the turn lane to the edge line past the turn. Detailed Design Report 11 August 10, 2001 Yellow Skip or One Solid Yellow Solid White Picture 2: Standard two lane Yellow Skip Solid White Black Picture 3: Two lane with center turns Detailed Design Report 12 August 10, 2001 Solid Yellow or Yellow Skip Solid White Picture 4: Three lane to two lane Black Solid White White Skip Picture 5: Four lane with center median and right turn Detailed Design Report 13 August 10, 2001 2.1.4 Tape Specification The “Preliminary Specifications Lane Awareness Tape” is attached as Appendix A. It lists all the appropriate specifications for the tape to be used by this project. 2.1.5 Tape Warranty There is a four year warranty on the magnetic tape including tape adhesion, reflectivity, and magnetic field strength. 3M will be responsible for tape inspection along the specified corridor for the duration of the project. Mn/DOT will be responsible for the tape inspection after the project completion. 3M will be responsible for all magnetic tape service necessary for the duration of the warranty period. 2.2 VISIBILITY SYSTEM 2.2.1 Need for Visibility Measurement The main focus of the Field Operational Test is to improve the ability of a driver to safely navigate and guide a vehicle under conditions of low visibility. In order to document and quantify precisely how well the driver assistive system works under these conditions, visibility conditions for the duration of the test must be quantified and recorded. This document describes a system capable of computing and recording visibility conditions as they relate to this field operational test. 2.2.2 Background In general, when the term “visibility” is referenced to describe weather conditions, it commonly means the scientific definition, meteorological visual range (MVR). According to the American Meteorological Society’s definition, visibility represents the maximum distance where a reasonable size of an object can be identified by a naked eye. On the other hand, MVR is measured based on the density of the atmospheric particles that affect the visible-spectrum transfer function of air space between two separated points where the measurements take place. Thus, MVR is consistent whether it is day or night as long as the density and distribution of the atmospheric particles are the same. MVR measurement does not consider any human factors such as human vision or environmental factors such as lighting conditions and obstruction of view by large objects. In this document, the term “meteorological visual range (MVR)” will be used whenever visibility refers to a weather condition. MVR has typically been measured using an instrument called a transmissometer that consists of a light source and telescopic photocell censors. However, due to its bulkiness and difficulty of installation, this instrument is considered unsuitable for transportation applications. Recently, a new measurement method based on a forward or backward light-scatter principle has been frequently used in highway and airport applications. This new measurement method facilitates a compact sensor design and easy installation for roadside applications. For the Highway 7 project, commercial MVR sensors developed Detailed Design Report 14 August 10, 2001 using the light scattering principle will be installed along the side of the road. In addition, for more through documentation of the driving conditions, additional weather sensors (i.e., wind direction, wind speed, humidity, and temperature) will be installed and record the data. MVR alone in general does not very well describe what visual conditions motorists actually experience on the road. For example, visual information is significantly reduced at night causing driving difficulty, but the MVR may show that the atmospheric visibility was very high. Similarly, driving against direct sunlight severely distorts the available visual information and creates a difficult driving experience, but MVR may indicate that it occurred during a very good visibility condition. It has also been shown by many researchers that, during a typical winter storm, atmospheric visibility varies significantly near the road surface due to the snow cloud generated by vehicles. In this project, the MVR sensors will capture the overall weather conditions along the highway, but in order to capture the visibility conditions experienced by the vehicle operators, additional invehicle sensors will be used. For this purpose, a color video camera is installed inside of each vehicle in a forward-looking direction to capture the images that the drivers were able to see on the road. Video images will be continuously recorded throughout the test period and digitally archived for the later review. Video images to be recorded provide the detailed view of what the driver experienced and thus critically important for the field-test documentation. However, due to a very large amount of data, it is very difficult to manually review each video image and analyze the driving conditions. As an innovative approach, a new indicator called the Motorist’s Relative Visual Range Index (MRVI) is introduced in this project, and the computational methods will be developed. MRVI is computed from the recorded video images using image processing techniques and quantifies how the visibility conditions local to the driver. The plot of MRVI will provide a quick view of how the visual conditions changed during a trip. Moreover, it will provide a means of quickly identifying a section of road that had low visibility, from which the reviewers can directly view the actual video images at that location. The computation of MRVI at a specified location is performed based on the loss of information relative to an image taken under an ideal weather condition (clear day with no obstruction of view). The range of MRVI will be between zero to one, representing the percentage of how close to the ideal condition a captured image is. MRVI will be zero when no visual information is available, e.g., a total white or black out condition. If the present condition is identical to the ideal condition, MRVI will become one. As visual information increases by the improvement of MVR or other conditions, this index will increase. In summary, MRVI should serve as a screening tool for the detailed video analysis of the road visibility conditions as well as for a quantifiable representation of local visibility. 2.2.3 Overall Visibility System The overall system consists of instrumentation suites that are required to measure the following two components: Detailed Design Report 15 August 10, 2001 1. meteorological atmospheric conditions (meteorological visual range, precipitation type/intensity, air temperature, humidity, wind direction/speed) 2. motorist’s view of the road condition (MRVI and video images). The first component is measured by the weather and visibility sensors that are implemented along the right of way of Trunk Highway 7. The second component is measured using the video images that are recorded using a video camera inside the vehicle looking in the forward direction. Figure 6 shows the overall block diagram. Digitized video data In-vehicle video camera MVR and weather sensor Highway 7 field operational test data archive MRVI computation Data acquisition Figure 6. Overall visibility system block diagram MVR and weather sensor system At six locations along Highway 7 between I-494 and Hutchinson, six new sensors capable of measuring MVR are to be installed. One weather sensor station is already available from the Mn/DOT R/WIS5 and is described later in this section. For the MVR, PWD11 developed by the Vaisala Co. will be used. The PWD11 is mounted on a mast installed on the right of way of Highway 7 and provides the following data: • • • MVR measurement in the range 10:2000 meters Precipitation type: precipitation, snow, rain, mixed Precipitation accumulation: ±30% in light and moderate rain, detection sensitivity of 0.10 mm/h (liquid precipitation) Of these six sensors, one will be augmented with additional sensors to provide additional information on atmospheric conditions. This particular sensor is denoted VW (for Visibility Weather); the other five sensors are denoted VO (for Visibility Only). The atmospheric conditions that will be measured from this station are: 5 Remote Weather Information System Detailed Design Report 16 August 10, 2001 • • • • Wind speed: maximum speed of 75 m/s Wind direction: 8 degrees or better resolution, 5 degrees or better accuracy Temperature: -40°F to +140°F, accuracy of +/- .5°F -40°F to +140°F Temperature or better Relative Humidity: 0 to 100%, accuracy of +/-5% One field station of the Mn/DOT’s Road Weather Information System (R/WIS) is presently located on the Field Operational Test corridor, and will complement the six additional sensors. This R/WIS station provides the following information: • • • • • • • • Visibility Wind speed Temperature Relative humidity Pavement condition (wet, snow, frost, ice, dry) Freezing point Precipitation type Precipitation rate The information coming from the R/WIS site is similar to that of the PWD11 station with the added weather sensors except that it includes the data on pavement conditions. Therefore, the seven sensor system will not only determine MVR at 7 discrete locations, but will also provide information regarding what weather conditions contributed to reduced motorist visibility. The overall system view is shown in Figure 7. VO VO TH-15 Hutchinson RWIS VO VO Highway 7 VW VO I494 Twin Cities Figure 7. Highway 7 visibility and weather sensor configuration: VO: PWD11 mounted on a pole mast, RWIS: Mn/DOT’s R/WIS tower, VW: PDW11 with additional weather sensors mounted on a pole mast. Detailed Design Report 17 August 10, 2001 Site East Coordinate North Coordinate Degrees Degrees Distance from Previous, Miles Side of Road H19 HW7 93.34.067 44.53.914 5.54 NW corner King PT Road 93.42.101 44.53.473 6.63 NE Corner M25 Hw7 93.52.136 44.54.341 8.3 SE Corner RWIS 7 93.59.147 44.54.380 5.79 South Side “Hot Tub” 94.07.138 44.54.4033 6.61 North Side Co. Road 90 94.13.619 44.53.613 5.53 North Side Omega Dr. 94.21.005 44.52.351 6.09 South Side Table 1. Location of Weather Stations along Highway 7 The locations described in Table 1 above were chosen based on a uniform distribution of weather stations along highway 7 and where right of way requirements can be met. Existing infrastructure has been considered as well. Only the locations at King Point road and at Highway 19 will require transformers to be installed at the site. Other locations can use existing power transformers as a source of electricity for these weather stations. Phone lines will be provided to these weather stations as well. A central data server (or servers, depending on long distance rules which may apply) will dial up each weather station on a 5 minute interval, and collected all available data from that weather stations. This software has been written, and will run under Windows NT, the operating system which will also operate the data server. Standard external US Robotics 56K modems (Model Number 839) are installed inside the weather station enclosures, and these pass data over the phone lines to the data server. All six weather stations have been configured in our lab, and test data has been collected using the University phone system. All that is left to do is install the weather stations along Highway 7. The weather station base design is shown in Figure 8. Each weather station to be installed will use a 24 inch diameter, 72 inch deep concrete base as shown. One to three inches of the foundation will be exposed above grade. Three anchor bolts are embedded in the concrete; the masts used to support the sensors are fastened to the foundation using these anchor bolts. A PVC conduit is also provided in each foundation. Power and telephone lines will be trenched from the respective source to the weather station foundation; the PVC conduit will allow an easy cable run to the weather station electronics. Construction will begin as soon as the ground thaws. Power and phone providers have been contacted, and visits to the sites have been made. As soon as the snow melts, a final Detailed Design Report 18 August 10, 2001 Anchor Bolts 24 “ Cable Conduit 60 “ 12 “ Figure 8. Weather station foundation design location meeting will be held with contractors, power and phone providers. As soon as final locations have been determined, contracts will be established, and work shall begin. As soon as the foundations are complete, power and phone lines will be run, and connections finalized. Once connections are made, the stations will go on line. 2.2.4 Video recording and MRVI computation system In each vehicle, a CCD color video camera is installed on the dashboard in a forward looking direction in order to capture images at a similar angle of view as the driver. The image capture board installed in the vehDAQ computer digitizes and compresses only the luminance of the video signal. The reason for using a color video camera but digitizing only the luminance is to record image data only in the visible spectrum band. Normal B/W CCD video cameras are sensitive slightly outside the color spectrum such as near infrared and near ultraviolet capturing the spectra that are not visible to human eye. The specification of the camera chosen for this project is as follows: • • • • • • • • Sensor: ½” color CCD Signal: NTSC IRIS: auto Focal Length: 12mm Max aperture: F1.4 Focusing: manual Lens mount: CS Manufacturer and model no: Cohu Inc., 1322-1000/EH13 Detailed Design Report 19 August 10, 2001 Each video image includes vehicle position and orientation information. This information is used to compute the local geometry, from which image segmentation is achieved. MRVI is then computed using the segmented image pieces. Since MRVI is only used as an analysis tool, the actual computation will performed off-line using a special computer allocated for image processing. An appropriate image processing software will be developed so that the computing process is completely automated. The software will generate the MRVI data as well as a graph of each run of the video data. A block diagram of the MVI computational system is given in Figure 9. digital image(s) of interest from vehDAQ forward looking camera Driver’s view out windshield MRVI Processor MRVI for image(s) of interest Local road geometry Engineering unit data synchronized with digital image(s) Vehicle position and orientation Geo-spatial database Figure 9. MRVI computation data flow diagram MRVI Computation Algorithm MRVI is intuitively defined as an index of motorist’s effective visual information that describes how clearly the road and surroundings are visually available to the motorist. It is assumed that what the motorist sees through the font windshield of a vehicle is similar to the video images captured through a forward-looking color camera. This is a gross simplification, since human vision is a stereoscopic system where 3-D information is available. Human vision is also known to have much higher resolution as well as multiple angles of view when compared to a fixed video camera. However, such differences may not be significant in the one-dimensional MRVI computation. A further assumption is made to make the computation tractable. It is assumed that MRVI can be computed from the luminance of the color images. Under above assumptions, MRVI computation is obtained using two data sources: • • Digitized luminance images generated from a color video camera looking through the windshield in a forward direction inside the test vehicle Local road geometry computed from the pre-recorded geo-spatial database. Detailed Design Report 20 August 10, 2001 MRVI Theory Visual range is defined by the maximum distance where a typical motorist can identify a reasonably sized object by the naked eye. Under low visibility conditions, the motorist’s effective visual information with respect to road geometry is assumed to be mainly affected by the atmospheric visual range. This is a reasonable assumption if psychological and the prior knowledge aspects of motorists are excluded. More specifically, as the motorist is able to see farther down the road ahead, the better he or she would be able to assess the driving condition. We will refer the visual range mentioned above as the motorist’s visual range in order to distinguish it from the term “visual range” used in atmospheric science. Direct computation of motorist’s visual range in terms of distance would be very difficult, since it is a function of many factors such as weather, geometry of road, lighting, etc. Therefore, we introduce an index referred to as the motorist’s visual-range index (MVI) that is simply computed by weighted summation of the influential factors. MVI measured at location l at time t is denoted as MVI(l,t). Now, suppose that we compute MVI using an image taken under an ideal visual condition, i.e., clear day with no cloud cover and no obstruction of view. Let this MVI be denoted as MVIopt(l), then MVI opt (l ) = Max MVI (l , t ) (1) t The Motorist’s Relative Visual-Range Index (MRVI) is then defined as MRVI (l , t ) = 1 − MVI opt (l ) − MVI (l , t ) (2) MVI opt (l ) Notice that the range of MRVI is [0,1]; the maximum occurs when MVIopt(l) =MVI(l,t); the minimum occurs when MVI(l,t). Essentially, MRVI represents how close the present visual road condition is to the ideal condition. Many factors may influence the motorist’s visual range. Three factors are considered most significant. The first factor is the atmospheric visibility (or MVR). It is mainly determined by the weather condition such as fog, rain, and snow and limits the motorist’s visual range. The second is the geographical factor. If the road ahead has a stiff uphill, motorists can only see up to the top of the hill and will not be able to see anything behind the top of the hill limiting the visual range. In general, downhill roads have a longer motorist’s visual range than uphill roads. A similar condition may occur on sharply curved roads or intersections where motorists have to make a turn and cannot see beyond the sharp curve or the intersection due to the blockage of view by buildings or trees. However, this factor is not considered as significant, since vehicles usually slow down in intersections or at sharp curves until the motorist is able to see the road in the traveling direction. The third factor is visually identifiable characteristics of the road boundaries. If Detailed Design Report 21 August 10, 2001 a motorist cannot distinguish where the road and lane boundaries are, it creates greater difficulties in navigation affecting the motorist’s effective visual range. Therefore, motorist’s visual range index at location l at time t is modeled as: MVI(l,t) = A(l,t) + G(l) + D(l,t) (3) where A is the atmospheric visibility factor, G is the geographical factor, and D is the road difference factor that determines the visibility of road boundary information. Note that the geographical factor is only a function of location; it is assumed that geographical alleviation changes of a road do not occur within the field operational test period. There will be other factors that may influence the motorist’s visual range such as clarity of the vehicle’s front windshield and luminance conditions. However, since A(l,t) is computed using the video images that are captured by a video camera looking through the front windshield, effects such as clarity of the windshield and brightness conditions will be in some degree included in the computation of MVI. There is some correlation between A(l,t) and D(l,t), e.g., as weather gets bad, it is more difficult to distinguish between where the road boundaries are. Substituting Eq. (3) to Eq (2) gives MRVI (l , t ) = 1 − Aopt (l ) − A(l , t ) + Dopt (l ) − D(l , t ) MVI opt (l ) (4) where Aopt(l) > A(l,t) and Dopt(l) > D(l,t). Notice that since G(l) is only a function of location, it does not contribute to the MRVI(l,t). Next we consider how to compute Aopt(l) from a digitized image. According to Duntly and many other researchers, visual range is affected by contrast. The basic principle that governs this relationship is called the Duntly’s law. MVI Computation from Video Images One of the most fundamental principles of visibility is expressed using luminance a black target as: Lb = L0b e −σx + Lh (1 − e −σx ) (5) where L0b is the inherent luminance of the black target that is non-zero, Lb is the apparent luminance, σ is the extinction coefficient, and x denotes the distance from the target. Assume that an object A has background B. Writing luminance equations for A and B and finding the difference gives: L A − LB = ( L0A − L0B )e −σx , Detailed Design Report (6) 22 August 10, 2001 where L A and LB are apparent luminances of A and B, and L0A and L0B are inherent luminances of A and B, respectively. Using differences, it is denoted as: 0 D AB = D AB e −σx (7) The difference D AB represents a contrast formed by foreground A and background B in an image. Unfortunately, in a general scene of image, the region of foreground and background is not obvious and difficult to recognize. One reasonable approach is computing the differences in all direction instead of computing error-prone object recognition. Let a 3x3 segmented regions of an image be denoted as: P1 P4 P7 P2 P5 P8 P3 P6 P9 where Pz denotes the pixel or representative region value at location z. Then, we define the point contrast in all directions computed at the location 5 as: C5 = 1 1 | ( P1 + 2 P2 + P3 ) − ( P7 + 2 P8 + P9 ) | + | ( P1 + 2 P4 + P7 ) − ( P3 + 2 P2 + P9 ) | 4 4 (8) Next, let Cz denote the apparent point contrast at a location x and g(z) be the distance computed from the observation point (i.e. camera) to the location z. It then follows the basic visual range relation in Eq. (7), C z = C z0 e −σg ( z ) , (9) where C z0 denotes the inherent point contrast. In our case, the real distance function g(.) is known because of the retrieval capability of the geospatial database based on the GPS coordinate of the observation point. Moreover, if visibility V is determined at a threshold ε = C z / C z0 , then the visual range is obtained from Eq. (9) as: V =− ln ε (10) σ Next, we choose a region of interest where C z0 is relatively constant over the long range from the observation point. Such an area typically exists in a highway in the tree line of the road. Suppose now that the extinction coefficient in this region is σ 1 , then the apparent point contrast must satisfy: C z = C z0 e −σ 1 g ( z ) Detailed Design Report (11) 23 August 10, 2001 and is depicted in Fig 10. C z0 g(z) Figure 10: Point contrast to distance relation Integrating the area below the curve (Eq. (11)) from distance zero to infinity gives: ∫ ∞ 0 C z0 e −σ 1 g ( z ) dg ( z ) = CZ0 (12) σ1 This relation provides how the extinction can be computed using video images and therefore computation of visual range using Eq. (10). From this analysis, we come to a conclusion that visual range can be computed by integrating (summation in the digitized data) from distance zero to infinity. For this project, our goal is computing MVI which represents more than the visual range, i.e., it should represent the motorists visual information, more sophisticated method are devised. The highway images are segmented into three regions a shown in Figure 11. Region 3 Region 1 Vehicle head Region 2 Figure 11. Regions of interest in the image In a highway scene, three regions shown in Figure 11 are considered most important. Regions 1 and 3 represent the road boundary and shoulder area and Region 2 represents the road area. The MVI form these separate regions is the computed as: Detailed Design Report 24 August 10, 2001 MVI = 1 ( w1C R1 + w2C R2 + w3C R3 ) , w1 + w2 + w3 (13) where wi denotes the weighting factor of region i that is determined based on the importance of the region of interest and C Ri denotes the average of the point contrasts in Region i. For the determination of the weighting factors, experiments must be conducted using real images. In general, the inherent point contrast of Region 2 varies depending on the number of vehicle present, such that it is much less stable (reliable) information than the information obtained from Region 1 or Region 3. Thus less weight should be given. In comparison of Region 1 and Region 3, Region 3 should have more importance due to right hand driving practice in US. Thus one simple weighting relation suggested (and used herein) is: 1 MVI = (2C R1 + C R2 + 3C R3 ) 6 (14) The average point contrasts computed in Eqs. (13) and (14) are based on pixel or segmented region levels. In order to generalize the relation and to maximally capture the information, we suggest it to be simply computed through multi-resolution of images, i.e., MVI = Max( MVIξ i ) , (15) i where MVIξ denotes the MVI computed at a resolution ξ i . Here different resolution is i equivalent to the decimation function in the wavelet space. This decimation computation will ensure that the capture of point contrasts is determined based on analysis from large to small objects and selects the highest quality of point contrast by choosing the maximum. In practice, two levels of decimation may be sufficient. For example, if the original image has the resolution of 640x480, MVIs are computed from 640x480, 320x280, and 160x140 images, and the maximum is selected. In summary, MRVI is computed using the following steps: 1. 2. 3. 4. 5. 6. 7. 8. 9. Read an image Median filtering of the original image to remove impulsive noise Create three different resolutions of the image Determine the three regions using spatial database for all three resolutions Compute MVI using Eqs. (8) and (14) for all three resolutions Select the maximum MVI, i.e., Eq. (15) and record it Compute MRVI using (2) and record it Update MVIopt using (1) Go to 1 Detailed Design Report 25 August 10, 2001 2.3 GPS BASE STATION INFRASTRUCTURE With GPS Selective Availability (SA) turned on, an uncorrected GPS measurement is accurate to approximately 100 meters Circular Error Probable (which roughly translates to one standard deviation in both North-South and East-West axes). With former President Clinton’s 01 May 2000 announcement that SA will be turned off as of midnight 01 May 2000, accuracy of uncorrected GPS approaches the 10 meter level. Although a drastic improvement, 10 meter position error is approximately 2 orders of magnitude greater than that necessary for vehicle navigation. For vehicle guidance applications, vehicle position has to be known to an accuracy of 20 cm or better. To obtain this accuracy for a civilian application, Differentially Corrected GPS (DGPS) is required. The following documents the requirements for the infrastructure for the DGPS system used in this Field Operational Test. 2.3.1 Background DGPS functions on the assumption that errors in the GPS position solution arise from delays caused in the propagation of radio signals as they pass through the Earth’s Ionosphere and Troposphere. If the base receiver is located a “short distance”6 from the roving receiver, the signal propagation delays experienced by both receivers are quite similar (i.e., common mode). Baselines are typically kept to distances of less than 10 Km, but University of Minnesota experiments done with Trimble equipment indicated accuracy is maintained over longer baselines (approximately 11 miles with an error of approximately 10 cm). Phone conversations with Leica indicate receiver accuracy is maintained with their equipment at baselines of 28 km. Typical carrier phase GPS receivers typically claim accuracy at 2 cm + 2 ppm, where the ppm is the baseline distance, for baselines less than 10km. For this project, the DGPS infrastructure consists of DGPS correction broadcast stations7, and high accuracy geo-spatial databases. The correction stations can be further reduced into the towers, equipment placement, radio electronics, and GPS antennae and receivers. As previously described, the DGPS correction signals provide the means with which to obtain high accuracy vehicle position solutions in real time to enable vehicle lateral and longitudinal navigation and guidance capabilities. The high accuracy geo-spatial database provides the roadway references used to maintain desired lane position. Although the digital geospatial database is carried on the vehicle and stored in computer memory, it is considered to be part of the DGPS system infrastructure in that it describes 6 7 The distance from the base station to the receiver is known as “baseline.” The in-vehicle GPS receivers are discussed in the TRS-Vehicle subsection . Detailed Design Report 26 August 10, 2001 From Rovers From Satellites Top of Tower GPS Base Antenna broadcast antenna Tower Base Lightening Protection GPS Base receiver Station Control Computer correction RF Modem Request for correction Figure 12. GPS correction base station configuration the presence, location, and physical attributes of relevant elements of the road infrastructure. A schematic of the correction hardware is provided above in figure 12. The base station operates on a “correction on demand” basis. The GPS receiver continuously receives signals from the satellites, and continuously computes corrections. These corrections, in the CMR (Compact Measurement Record) format, and sent to the base station control computer over an RS-232 serial line. The control computer buffers these corrections until they are needed. The corrections are needed when a vehicle in the correction zone (to be addressed below) requests a correction from the tower. The RF Modem in the tower will receive this request for correction, and will pass it to the control computer via the RS 232 line connecting the two. The control computer holds the most recent GPS correction, adds header and footer information, and at the proper time, commands the RF modem to broadcast the correction at the proper time, and then commands the RF modem to go silent until the next correction is sent. As long as a correction has been requested, this process is repeated once per second. Detailed Design Report 27 August 10, 2001 The equipment used to execute this base station control is provided in table 2 below. Component Manufacturer/ Model Comments GPS Antenna Leica ST 502 No choke ring needed Lightening Protection Polyphasor IS-50NX-C2 2 locations GPS Receiver Leica SR 530 10 Hz, CMR compatible Control Computer 386 processor, 8 MB Disk on chip Broadcast Antenna Emac GLB Wireless SNRDS, 453.6375 MHz Celwave PD1150-6 GPS Cable Leica 664813 Std. Leica Cable RF Modem Cable Andrew LDF4.5-50 Low Loss 5/8” coaxial cable RF Modem 25 Watts, Field Adjustable Power 5 db gain, omnidirectional Table 2. Base station equipment list Base station timing is required so the entire test corridor can be covered with a signal radio frequency. In the simple, channel hogging approach, three frequencies (one for each base station) would be required to broadcast a correction from each base station. This solution is less than optimal, however, because it is expensive (three channel RF modems would be needed in each vehicle) and it hogs precious radio spectra (three bands). In the approach used for this project, only one frequency is needed to broadcast the correction along the test corridor. Single frequency broadcasts are possible because of the structure of the CMR correction and the timing information available from GPS. The CMR message has a maximum length of approximately 2200 bits; GPS timing information is precise to a level better than the microsecond. The FCC license issued for this project is a 12.5 KHz band centered at 453.6375 MHz. The 12.5 KHz band will support 9600 baud communications. A 2200 bit message can be broadcast in 229 milliseconds. This allows for the complete broadcast of the CMR message and header information in a 0.5 second interval. The approach then is that the Chanhassen and Silver Lake stations broadcast on the 0-to-0.5 second interval; the Mayer station will broadcast on the 0.5-to-1.0 second interval. This is illustrated below in figures 13 and 14. Detailed Design Report 28 August 10, 2001 Figure 13. Corridor showing locations of GPS base stations. Green is Silver Lake, blue is Mayer, and red is Chanhassen. On Off 0.5 1.0 1.5 2.0 Time, seconds Figure 14. Timing of GPS base stations along Highway 7. Eastern and western stations are active while center is quiet, and vice versa. Detailed Design Report 29 August 10, 2001 Figure 15. RF Signal propagation for GPS base stations. Coverage is complete throughout test corridor. The RF modems used for this project will operate at 453.6375 MHz. Figure 15 above shows the propagation study results showing the coverage of the three radios at an effective radiated power of 25 watts. Because a 3 db gain antenna has been chosen, the base stations radios will operate at a power level at less than 12 watts. The modems are field adjustable, so if coverage is weak anywhere, base station power can be increased so that full coverage is assured. The reliability of the propagation analysis is a function, among other variables, the model of the terrain surrounding the GPS base stations. The GPS receivers, GPS antenna, and the RF modems used in the base station are the same as those used on the vehicles. Therefore, to avoid duplication, descriptions of these components are deferred until the vehicle section. At the time that this draft was prepared, lease agreements had not yet been finalized with each of the three cities. Both Silver Lake and Two Way Communications have agreed to terms; all that remains is the final signature. The City of Mayer has also agreed to the terms; however, Carver county, which has an equipment shed at the base of the tower, has agreed to allow the University to place GPS equipment in the shed; the no-cost lease agreement is still being negotiated. The agreement with the city of Chanhassen has been signed as of 04 April 2001. Contracts for installation with International FiberCom, a local tower contractor, have been in place since fall 2000. As soon as agreements are finalized with the cities, installation will commence. All material required for radio installation has been purchased and is currently in storage at the University of Minnesota. Installation of the radio equipment will be straightforward. The only impediment will be the installation of a cable port on the Chanhassen tower. This tower, has no previous RF equipment located on it. As such, a welder will be needed to install a cable run at the top Detailed Design Report 30 August 10, 2001 of the tower. This makes installation a two-day job. Otherwise, the other sites should be up and running in less than one day. The system will be verified in two different manners. First, GPS base stations need to provide seamless coverage along the test corridor. This can be verified by travelling from one end of the corridor to the other to document that GPS corrections are provided throughout the corridor. Second, each vehicle has to “know” from which base station to take the correction. This can be verified by collecting the GPS correction data taken as a vehicle is driven from one corridor to the other, and then looking to see if the correction has switched as scheduled. Third, the accuracy of the GPS system has to shown to meet spec. This will be done both statically and dynamically. Static testing consists of colocating the GPS antennae with a known accurate reference. Data will be collected at those sites, and position solutions computed and compared to the reference. Once static accuracy is known, dynamic accuracy can be tested in a number of ways. Aggregate system (including GPS) accuracy can be determined using images captured by a camera “viewing” the HUD, or GPS accuracy alone can be determined by using a calibrated camera synchronized with GPS to capture images as the camera passes over ground located references. The location of the GPS antenna is known with respect to the camera; the camera/lens calibration will indicate where the reference is with respect to the camera. From that, the physical GPS antenna phase center location can be compared to that of the ground located reference. Photos of the towers on which the base stations will be located are provided on the following pages. Detailed Design Report 31 August 10, 2001 Picture 6: Chanhassen Melody Hill Water Tower Detailed Design Report 32 August 10, 2001 Picture 7: Mayer Water Tower Detailed Design Report 33 August 10, 2001 Picture 8: Silver Lake Water Tower Detailed Design Report 34 August 10, 2001 2.4 GEOSPATIAL DATABASES The high accuracy geospatial database provides the roadway references used to maintain desired lane position, provides an ability to change lanes, and provides a means to filter from the vehicle operator unwanted radar returns from stationary elements in the geospatial landscape. Although the digital geospatial database is carried on the vehicle and stored in computer memory, it is considered to be part of the DGPS system infrastructure in that it describes the presence, location, and physical attributes of relevant elements of the road infrastructure. The structure and query process involving this geospatial database are considered proprietary, and have been submitted for consideration for a U.S. patent. As such, details will not be provided unless confidentiality can be assured. The University of Minnesota is presently involved with research involving the rapid, economical construction of these geospatial databases. Some of the technology developed therein will be applied to the construction of the databases used for this project. For this project, a video image capture system will be installed on the front of the SAFEPLOW research vehicle. As the SAFEPLOW travels Trunk Highway 7 and County roads 7 and 4 in McLeod County, a calibrated camera/lens system will capture images at 10 Hz. These images will be synchronized with the GPS system on the SAFEPLOW. Knowledge of the camera/lens calibration and the location of the camera with respect to the GPS antenna located on the SAFEPLOW allows for precise positioning of the paint stripes. Cataloging of road furniture local to Trunk Highway 7 will be done by discrete mapping using a mobile GPS antenna and a long cable attached to the GPS receiver in the SAFEPLOW. This road furniture will be cataloged by location and included in the database will be relevant attributes. For example, a “stop ahead” sign is located as a “sign” and cataloged with the attribute “stop ahead.” The Geospatial Database will be verified in two ways. First, locations of geospatial elements along highway 7 (a particular skip line, for instance) will be measured with a GPS receiver8 to determine its position and attributes. That location will then be queried, and the returned results compared to the surveyed point(s). The database search had better return the same information as is known physically by that surveyed point. The second verification step is to illustrate that the database and query engine process data requests at a rate sufficient for vehicle operation. This will be done by operating vehicles at speed, and observing the HUD for consistency of data and excessive errors. If neither occurs, the database is obviously performing to spec. 8 The accuracy of which will have already been documented. Detailed Design Report 35 August 10, 2001 Detailed Design Report 36 August 10, 2001 3.0 VEHICLE DESIGN 3.1 OVERVIEW The vehicle system can be thought of as the integration of four specific subsystems: the guidance system, the collision avoidance system, the computational system, and the driver interface system (aka human – machine interface). These are all clearly identified in the system overview in Figure 1 located in Chapter 1. The block diagram is repeated below in Figure 16 for convenience. Inertial Measurement Unit acceleration, rotation rates Driver Visual Interface position Magnetic Sensor lane lateral position Guidance Processor Geospatial Database Processor local landscape orientation Satellite Signals GPS receiver GPS Corrections vehicle global position forward sensors Driver Haptic Interface Driver Interface Processor Radar Processor relevant obstacles Driver Audible Interface side sensors rear sensors Figure 16. Vehicle System Block Diagram Herein, the design of each of the four subsystems is described in significant detail. Component design is not described, but component function (and specifications where necessary) are provided. Detailed Design Report 37 August 10, 2001 From correction antenna From satellites RF modem antenna GPS Antenna correction Control Computer RF Modem GPS Receiver global position request for correction Inertial Measurement Unit (IMU) 3 axes acceleration Navigation Process 3 axes rotation rate Magnetic Sensor Lateral Displacement from Tape Vehicle Position, orientation, displacement from tape Blue RS 232 Red RS 485 Figure 17. GPS Vehicle block diagram 3.1.1 Vehicle Guidance Subsystem The guidance subsystem consists itself of three subsystems: the GPS/IMU subsystem, the Inertial measurement subsystem, and the guidance processor. A description of how the system operates is provided first; components are described thereafter; the block diagram is shown in Figure 17 above. Detailed Design Report 38 August 10, 2001 GPS is used as the primary positioning sensor. When GPS satellites are in the field of view of the antenna and the distance to the base station is less than 10 miles9, the GPS system is capable of providing position estimates accurate to 2 cm at a rate of 10 Hz. A single GPS receiver and antenna is capable of providing vehicle position, velocity, and heading estimate information to the positioning process; heading angle estimates are provided by successive GPS solutions. The resultant heading vector is shown below in Figure 18. Position at T1 Position at T0 Figure 18: Heading estimate from successive GPS position solutions At relatively low speeds (less than 30 MPH), the GPS system provides position and heading information at an accuracy sufficient for safe vehicle navigation. For instance, at 30 MPH, a vehicle travels 44 feet per second, or 4.4 feet per GPS solution. During the 0.1 second the vehicle operates without a new GPS position solution, uncertainty regarding the position of the vehicle along its longitudinal axis is present.10 On the average, the error for this 0.1 second period is 2.2 feet. This has not proven to be a problem for low speed operation. See image below (Picture 9). However, at higher speeds (those anticipated by law enforcement and ambulances, for instance), this average error can increase to 4.4 feet (for a vehicle travelling 60 MPH). To account (and correct) for this error, an IMU is integrated with the GPS solution. Although the details are proprietary, the concept is straightforward. The IMU provide vehicle frame of reference longitudinal, lateral, and vertical accelerations and rotational rates along the vehicle longitudinal, lateral, and vertical axes. These accelerations and rotational rates are used to drive the Euler equations which describe the motion of a body through space. The idea behind the IMU/GPS integration is that the Euler equations of motion will be mathematically integrated at a rate an order of magnitude greater than the GPS solution rate. The integration uses the GPS position and velocity as initial conditions for the integration. By doing so, vehicle position and orientation estimates are provided at a much higher rate than position estimates from GPS alone, thereby eliminating much of the errors which arise because of the 0.1 sec. period between GPS solutions. This technique substantially improves system accuracy and performance. 9 Longer baselines are possible, but for the sake of brevity, this constraint will be imposed. Lateral uncertainty in the vehicle frame of reference is considerably less because of low relative lateral velocities. 10 Detailed Design Report 39 August 10, 2001 One issue which arises with GPS is that the satellite signals can be blocked by tunnels, overpasses, tall buildings, etc11. During the time the satellite signals are blocked, and the time it takes to recover (typically 10-15 seconds with high end receivers), the IMU will continue to calculate estimates of the vehicle position and orientation through the integration of the Euler equations of motion. Although the quality of IMU sensors has improved drastically over the past 5 years, these sensors still suffer from bias and bias drift. The bias and drift affect sensor accuracy, and therefore the accuracy of the solution calculated using the inputs to the IMU. The accurate GPS solutions are used to compute and correct for the bias and bias drift errors (in real time) coming from the IMU measurements. However, even with this correction, an affordable IMU can only provide sufficient accuracy to support the driver assistive system for approximately 30 seconds. The sensitivity of the driver assistive system to this error is greatest in the lateral axis of the vehicle. To extend this time for which the driver assistive system can operate without GPS, an additional reference to correct for lateral vehicle position error can be used. Magnetic tape embedded in the roadway can be used to provide a lateral position reference when the GPS signal is unavailable. This lateral reference can be used to correct for IMU sensor bias and bias drift errors, but because only one axis of information is available, the degree to which bias errors are corrected is somewhat limited. In operation, with the loss of GPS, the IMU alone can be used to provide sufficiently accurate lateral and longitudinal global information to operate the driver assistive system in its intended state (giving longitudinal (preview) information as well as lateral information). This time period is approximately 30 seconds. After this time, if no magnetic tape is available, the driver interface informs the driver that assistance is no longer possible, and the system reverts to “stand by” until GPS lock is reacquired. If the magnetic reference system is available, the position processor can provide sufficiently accurate information for approximately one additional minute to operate the driver assistive system in its intended state. Once global position estimates are no longer sufficiently accurate and the magnetic reference is still available, the driver assistive system then supplies only lateral information to the driver via the normal channels (HUD, tactile, and audible feedback). If GPS lock is reacquired, the system returns to its normal mode; if the vehicle operates beyond where the tape is located, the driver is informed and the driver assistive system reverts to “stand by” until either GPS returns or the vehicle is driven again over the tape, at which time the appropriate interface will be provided. 11 Foliage can be a problem if it directly overhead; however, if trees are trimmed along a highway, problems rarely arise with high end receivers. Detailed Design Report 40 August 10, 2001 Picture 9: Head Up Display Image captured at 30 MPH on Highway 101 Northbound between Rogers and Elk River, Minnesota. Note fidelity with which lane boundaries are projected into the HUD. GPS / IMU Subsystem Specifics The GPS subsystem consists of a GPS receiver and antenna, a RF modem used to receive corrections from a base station, and an interface to a data acquisition or control computer. A block diagram of that subsystem was provided in Figure 17. The component list for the GPS / IMU subsystem is provided in Table 3 below. Item Vendor/Model Comments GPS receiver GPS antenna RF Modem RF Antenna IMU Leica SR 530 Leica AT 502 GLB SNRDS Antenex Phantom Crossbow HDX 10 Hz, CMR correction Dual Frequency 25 Watt 453.6375 MHz, 0 db gain 3 axes accel, 3 rotation rates Table 3. GPS Vehicle Parts list Detailed Design Report 41 August 10, 2001 GPS interaction between vehicle and base station The system works in that as soon as the driver assistive system is turned on, the vehicle control computer requests a GPS correction from the nearest GPS tower. This request is broadcast from the vehicle to the tower. When the request is detected by the tower control computer, the tower begins to broadcast its correction in the specified time slot. Once the tower broadcasts its correction, the correction is received by the RF modem and passed to the vehicle control computer. The control computer separates the header from the CMR correction, and passes it along to the GPS receiver. It may be such that the vehicle in question is between two GPS correction “zones.” The RF modem will be receiving corrections from both GPS base stations, each correction arriving in the appropriate time slot. As a redundancy, each modem is assigned an ID which is included as part of the broadcast correction message. The vehicle control computer uses the on-board geospatial database to determine which GPS base station is closest to the vehicle. Incorporated in the database is the station ID which is part of the broadcast string. The desired GPS correction message is identified as such when the desired ID matches that broadcast by the GPS base station. GPS Receiver. The requirements for the GPS receiver to be used in this project were defined in the TSR and are repeated below for convenience. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. Accuracy: 5 cm + 5 ppm horizontal; 10 cm + 10 ppm vertical Receiver channels: 9 minimum Cold Start Initialization: less than 2 minutes Reacquisition time from recent epoch: L1 < 5 seconds, L2 < 30 seconds True (not estimated) position solution rate: 5 Hz minimum Solution latency: < 50mS Correction accepted: CMR required, RTCM 18,19,20,21 desirable Corrections broadcast: CMR required, RTCM 18,19,20,21 desirable Temperature operating: -20 degrees C to 60 degrees C Vibration: 3 G operating, 6 G surviving Mechanical Shock: +/- 30 G Operating, +/- 60 G surviving Solution and correction communication: RS 232 Based on cost: performance considerations, the Leica SR 530 GPS system was selected for both the vehicles and the base station. Illustrations of the GPS/IMU system are provided in Pictures 10-15 on the next pages. Detailed Design Report 42 August 10, 2001 Picture 10: Crossbow HDX IMU Removable Pendant (not used in vehicles) GPS Receiver Picture 11: Leica SR530 GPS Receiver. Red user interface (pendant) is removable. Only the white box will be located in the vehicles and in the base station. Detailed Design Report 43 August 10, 2001 Flash Memory Port. Pulse per second marker Picture 12: Leica SR530 GPS receiver, Front View Threaded insert for mounting stud. Picture 13: Leica AT 502 Dual Frequency GPS antenna Detailed Design Report 44 August 10, 2001 GPS corrections are received from the tower by the vehicle using standard off-the-shelf radio data modems. The antenna used for corrections is shown in Picture 14 below; magentic or trunk lid / rain gutter mounts are available. The choice of mount will be decided upon when options are discussed with the respective vehicle owners. Picture 14: Antenex Phantom RF antenna, 453.6375 MHz. Magnetic mount shown. The data radio used for both the base stations and the vehicle are GLB SNRDS radios. An FCC license was granted for the support of these GPS correction for the center frequency of 453.6375 MHz. These modems support local networks, and can be configured with modem IDs. More information can be found at http://www.aria-glb.com. Detailed Design Report 45 August 10, 2001 Picture 15: GLB SNRDS RF Modem, 453.6375 MHz center frequency, 12.5KHz bandwidth. IMU and integration with DGPS For this project, the minimum specifications for the IMU are as follows: 1. 2. 3. 4. 5. 6. 7. 8. Roll, Pitch Angle: Dynamic Accuracy: 1° RMS Resolution: 0.1° Linearity: < 1 % of FS Full Scale Span (analog outputs): ± 4.096 VDC Full Scale Measurement ranges: 90 degrees full scale roll and pitch Operating temperature range: -20 to 60 degrees C Mechanical shock: +/- 30 G Operating, +/- 60 G surviving Mechanical vibration: 3 G operating, 6 G surviving As described previously, the Crossbow HDX six axis IMU was selected for this project based on both performance and price considerations. Detailed Design Report 46 August 10, 2001 Crossbow IMU Detailed Specifications: 1. Attitude A. Input Range i. Roll ± 180 deg ii. Pitch ± 90 deg B. Accuracy i. Static < ± 0.5 Measured on level surface ii. Dynamic ± 2.0 deg rms iii. Resolution < 0.1 deg 2. Angular Rate A. Input Range i. Roll, Pitch, Yaw ± 200 deg/sec B. Accuracy i. Bias: Roll/Pitch/Yaw < ± 1.0 deg/sec Scaled sensor mode ii. Bias: Roll/Pitch < ± 0.05 deg/sec Angle mode iii. Scale Factor < 1 % iv. Non-linearity < 0.3 % FS v. Resolution < 0.05 deg/sec C. Bandwidth i. > Hz -3 dB point D. Random Walk i. < 1.7 deg/ hr 3. Acceleration A. Input Range i. X/Y/Z ± 10 g B. Accuracy i. Bias: X/Y/Z < 8.5 mg ii. Scale Factor < 1 % iii. Non-linearity < 1 % FS iv. Resolution < 1.5 mg C. Bandwidth i. > 10 Hz -3 dB point D. Random Walk i. < 0.5 m/s/ hr 4. Data Output A. Digital i. Format RS-232 See “Digital Data Format” ii. Update Rate > 75 Hz Continuous update mode B. Analog 1 i. Range 0 to 5.0 VDC Detailed Design Report 47 August 10, 2001 Clearly, from the specifications, bias errors are still an impediment to the direct use of an inexpensive IMU as a stand-alone navigation device for a highway vehicle. For instance, the maximum specified bias error for this sensor is less than 8.5mg, or 0.0833 m/s2. This level of bias error would result in a 1 m error in vehicle lateral position in fewer than 24 seconds if this were the only error present in the IMU. Under actual operating conditions, this 1m error would occur in less time. To combat the inherent problem with sensor bias and bias drift, the accurate DGPS positions are used, in real time, to compute and then correct for bias errors. A simplified block diagram of the integrated GPS/IMU system is provided in Figure 19. Figure 19 requires some explanation. The black signal / data flow lines on the figure represent data available at the high rate (150 Hz) from the IMU. Blue data flow lines represent the positional (and derivative) data available from the GPS receiver (at 10 Hz). IMU data is collected and represents rotational rates and accelerations in the vehicle frame of reference. However, GPS provides position information in the global frame of reference. Using orientation data, vehicle reference frame data is transformed into the global frame via transformation T-1. After transformation into the global frame, accelerations are integrated twice (using position and velocity initial conditions derived from GPS data) to get global position, and the rotational rates are used to determine vehicle orientation. Updated vehicle orientations are fed into the transformation matrices, and the integration continues until a new GPS position is determined. Upon the arrival of the new GPS position, the integrators are reset, and the process repeats. The GPS system also is used to correct for bias and bias drift errors coming inherent with the IMU sensor. When a new GPS position arrives, that position is synchronized with the global position computed estimated by integrating IMU information. Assuming that the error which arises is due to sensor bias, the correction block computes that bias for that 0.1 second time interval; that bias is transformed back to the vehicle reference frame, and the subtracted from the raw IMU measurement. A similar approach is used to correct for rotational rate sensor bias errors, the main difference being that IC’s for the rate integrator come from the GPS estimated heading and the attitude estimate from the IMU. Detailed Design Report 48 August 10, 2001 global position GPS Rcrv global velocity accels From IMU Heading Estimate rot. rates + IC’s accels - vehicle frame of reference T-1 rot. rates INT INT Global Position Orientation INT global frame of reference IC’s IC’s attitude (roll, pitch) Accel Bias Error Computation + T Heading Bias Error Computation + INT Integrator Figure 19. Simplified block diagram of the GPS/IMU integrated system Detailed Design Report - 49 August 10, 2001 The process changes slightly when GPS is unavailable but the magnetic reference is available. In this case, the GPS signal is replaced by the magnetic sensor signal. Because the magnetic sensor only provides a lateral measurement (in the vehicle frame), only bias in the lateral vehicle accelerometer can be estimated using this technique. Thus, the equations are modified to reflect this limitation in information. The magnetic sensing system is described in further detail below. Magnetic sensor subsystem 3M is the manufacturer of the sensor bar. The sensor bar contains all the sensing and processing electronics. It is mounted to the vehicle with mounting hardware specifically designed for the specific vehicle type. Performance specifications are outlined in the “Preliminary Product Bulletin for Snow Removal Operations” included as Appendix A. This sensor hardware design will be used on all vehicle platforms unless specific modification are found necessary through the testing phase of the project. The sensor bar is shown in Picture 16 below. Picture 16: Magnetic sensor bar and connecting cable. Sensing electronics are all housed within the sensor bar. Key specifications regarding magnetic sensor system performance are provided below; for more detailed information can be found in Appendix A. The cable is specifically designed to withstand the harsh snowplow environment and limit EMI emissions generated within the sensor electronics. It attaches to the output of the sensor bar and transports system information through the vehicle fire wall into the cab area where it terminates at the junction box. The junction box allows the sensor data to be Detailed Design Report 50 August 10, 2001 accessed by the on board computer systems for further processing. The junction box is shown in Picture 17. 1. Optimal Vehicle Speed: Optimal operation at vehicle speeds ranging from 8 to 80 kph (5-50 mph). 2. Initial Signal Acquisition Distance: Operational after detecting the presence of the tape for four meters (13 ft.) when within detection range (distance from tape less than 90 cm (3 ft.). 3. Output: calculates distance from the center of the sensing electronics to the center of the tape. 4. Detection Range: ±90 cm (±3.1 ft.) from the center of the magnetic tape (validated in laboratory tests). 5. Distance Error: ±1 cm above the tape and ±5 cm at ±90 cm (±2 in. at ±3.1 ft.) from the center of the tape at speeds from 8-40 kph (5-25 mph) (validated in laboratory tests). 6. Operational Height: 15cm to 45 cm (6 in. to 18 in.) above the tape with optimal performance at 20 cm (12 in.). 7. Dynamic Height Changes: Detection range and accuracy are independent of dynamic changes in height. 8. Distance Update Frequency: 0.5 sec at 16 kph (10 mph) and 0.1 seconds at 80 kph (50 mph). 9. Operating Environment: –40º C to 80º C (-40º F to 176º F) and 10 to 99% humidity (non-condensing) over ambient temperature range. The Junction Box Hardware can accept inputs from two separate sensor bar cables. The majority of the cabling shown in this picture will not be used on this project but is currently apart of 3M‘s commercial product. The only cabling required for this application will be include speed, power, and sensor bar inputs. The Junction Box Hardware is shown in Picture 17. Detailed Design Report 51 August 10, 2001 Power Inputs Sensor bar inputs RS 485 Picture 17: Hardware Interface Finally, when both GPS and magnetic tape signals are present, each will be a reference against which the other can be measured. As the field operational test progresses, both GPS and magnetic tape data will be recorded (by the vehDAQ). Vehicle operators will be notified of any discrepancies through the system information interface; 3M and University of Minnesota personnel will be notified of any discrepancy, and will be responsible for fixing the problem. 3.1.2 Radar based Collision warning/sensing system There are three subsystems which comprise the collision warning/sensor system: the forward-looking sensor subsystem, the side-looking sensor subsystem, and the rearlooking sensor subsystem. All vehicles are equipped with the forward and side-looking systems; only the snowplows are equipped with the rear looking sensor system. This is because snowplows operate at speeds considerably lower than those of the general public, and as such are at a higher risk of being struck from the rear. Forward looking radar The minimum specifications for the forward looking system are as follows: 1. Minimum horizontal beam sweep: +/- 6 degrees from vehicle longitudinal axis. 2. Minimum number of obstacles in field of view that can be tracked: 8 3. Minimum Operating range: 10-350 feet 4. Minimum host vehicle speed range: 1.2 to 120 MPH 5. Minimum closing vehicle rate range: 0.5 to 100 MPH Detailed Design Report 52 August 10, 2001 6. Minimum azimuth resolution: 0.2 degrees 7. Maximum range error: 5% of reported range 8. Sensor should provide target information at a rate of at least 10 Hz. These sensor requirements are met by the Eaton Vorad EVT-300 radar sensor. The University of Minnesota has considerable experience with the EVT-300. The EVT-300 radar provides not only range and range rate information, but also provides azimuth angle information to multiple targets. Using this information and the location and orientation of the host vehicle provided by DGPS/IMU based vehicle navigation processor, and a geospatial database of the road, the target vehicles can be located with respect to the road [Newstrom, 2000]. This allows the filtering of radar returns which do not correspond to roadway vehicles and are thus not a threat. This effectively increases the signal to noise ratio of the radar system as perceived by the driver. When not harassed by false alarms, the driver is more accepting of these systems. Figure 20 shows the advanced radar processor implemented on the SAFEPLOW. The plow was driven southbound on Trunk Highway 101 which is northwest of the Twin Cities. The radar detected both valid targets (red x) and reflections from roadside furniture (blue circles). The radar processor was able to filter out the false returns (radar returns emanating outside the driveable road surface) while presenting the driver with a real target vehicle shown in red. For additional information, please see the paper on this subject which has been submitted to the 4th International IEEE Conference on ITS (acceptance pending). Because the ability to process radar using a vehicle location system and a geospatial database is unique to the University of Minnesota, radar manufacturers are forced to deal with the false target problem in other ways. The most common way to eliminate false alarms from roadside targets is to keep the width of the radar narrow. By keeping the beam narrow, radar returns from elements outside the lane directly ahead are kept to a minimum. Additional processing consists of using vehicle speed in the signal processing algorithm; if the radar detects what appears to be a stationary object, a warning is not sent to the driver. Under most, conditions, this is not a problem, but in some conditions, it is a substantial shortcoming. Detailed Design Report 53 August 10, 2001 5 SAFEPLOW on Hwy 101 South Bound x 10 3.469 Road Shoulder Road Island SAFEPLOW Path Valid Radar Target Filtered Radar Target 3.4688 North, Y (meters) 3.4686 3.4684 3.4682 3.468 8.3511 8.3512 8.3513 8.3514 8.3515 East, X (meters) 8.3516 8.3517 8.3518 5 x 10 Figure 20. Graphical representation of a geospatial database used for radar processing to eliminate false targets from elements not on a drivable surface (in this case, the elements are turn signal poles (see blue marks) located inside the traffic islands. The geospatial database allows the process to distinguish between fixed geospatial objects and, for instance, a stalled vehicle on the shoulder near a sign. Given the angular resolution of the radar sensor, the radar processor can use the geospatial database to identify the sign as a piece of road furniture, and the stalled vehicle as legitimate target. The driver, therefore, is warned that the stalled vehicle is there and poses a threat, but is not warned about the sign because it does not pose a threat. Because of the ability to process using the geospatial database, the beam sweep of 12 degrees for the EVT – 300 is somewhat limiting. Moreover, the field of view of the HUD is 30 degrees, which is wider than that of a single radar unit. Because it is obviously unsafe to present road information in the HUD, but not obstacle data, the field of view of the forward looking radar needs to be increased. Eaton-Vorad is not (yet) going to design a new radar for this application, so alternative needs to be implemented. The solution to the field of view problem is simply to array the radar so that a wider field of view is covered. With a 12 degree field of view for each vehicle, three co-located sensors can seamlessly cover approximately a 36 degree field of view. The approach taken here is a slight variation on the co-located sensor approach. Instead, the approach spreads the sensors out; two sensors are located at the sides of the vehicle, and at least one is located in the center of the vehicle. For snowplow, four radar are used: to are located in the center of the vehicle; one up high, and one on the grill. This dual Detailed Design Report 54 August 10, 2001 approach provides robustness, error checking, and fault tolerance. Each of the center radar is used to monitor the health of the other. Redundancy is also provided by the side radar; beam overlap allows both radar to cover a portion of the view in front of the vehicle; if an object is in the (a priori) common region, if both radar are healthy, both should indicate the presence and location of the common target. If both don’t report that common object, a problem lies with one. Figure 21. Solid model results of the snowplow radar configuration study. Note three radar arrayed on vehicle roof; fourth radar located on grill. To get the proper orientation of these arrayed radar, solid models of the SAFEPLOW (which is representative of the snowplows in the Mn/DOT fleet) and the individual radar units was created to check a variety of radar sensor configurations. Co-location was tested, as were a variety of other configurations. As it turns out, spreading the sensors apart across the top of the vehicle was found to be an optimal solution for snowplows. Vehicles without the large metal blade in front are better served by a similar array, but one which is mounted closer to the front bumper. Eaton Vorad radar are designed to be mounted close to the pavement. Figure 21 above illustrates the results of the solid modeling effort. This configuration allows the radar and the HUD to cover the same forward field of view. Eaton specifies a 7.3 volt input voltage for their sensors, and these sensors communicated via the SAE J1708 standard. To facilitate easy integration into the test vehicles, all of which have 12 volt power buses and are equipped with computers which have available RS 232 serial interfaces, conversions from 12 to 7.3 volts and from RS 232 to J1708 are necessary to integrate the sensors into the driver assistive system. Pictures 18 – 20 on the following pages illustrate the hardware necessary to make the conversions. Detailed Design Report 55 August 10, 2001 Computer interface J1708 to RS 232 converter Power connector (XLR) Interface to EVT 300 Radar 12V to 7.3V regulator Picture 18: Box Top. Radar power and communication interface box. RS 232 to control computer 12 V. Power Connector Picture 19: Front end of radar power and communication interface box Detailed Design Report 56 August 10, 2001 Connector to EVT 300 Radar Picture 20: Back end of radar power and communication interface box Side Looking Radar The side sensors consist of four 10GHz radar sensors (for snow plows) or two 10GHz radar sensors (ambulance and patrol car). The sensors are manufactured by Altra Technologies, and are the Proximity Detectors. A Proximity Detector is shown in Picture 21 below. . Picture 21: Altra Technology’s Proximity Detector for side object detection Detailed Design Report 57 August 10, 2001 The specifications for the Altra Proximity Detector follow: 1. Horizontal beam width: +/- 40 degrees 2. Operating range: 0 to 25 feet 3. Nominal range gates: Close alarm: 0 to 4 feet Far alarm: 4 to 12 feet A DOS hardware utility program can be provided which allows the range gates to be scaled. The interface between the sensors and the vehicle computer is an RS-485 serial bus. All sensors share a single bus. The general message information on this bus follows: Baud rate = 38,400 bps No parity 1 start bit 1 stop bit 7 data bits Message rate is: minimum of one message every 2 seconds; when alarm condition remains the same, a message is sent every 2 seconds; whenever alarm condition changes up or down message is immediate. Each message from the sensor to the vehicle computer is normally followed by a reply (acknowledge) message from the Radar Processor to the sensor. The acknowledge message is 4 bytes: 62,25,0,39. Other messages are possible, depending on conditions. These messages have other message numbers (the 1st byte) and can be ignored. Messages are distinguished from each other by an inter-message dead time of at least 5 character times. Rear Looking Radar Rear looking collision warning sensing is provided by an independent system, the Altra Technologies Rear Guard System. This system will only be provided on the four snowplows. The system consists of a single 76.5 GHz radar sensor and two ultra-bright white flashers which face the rear. When the sensor picks up a vehicle approaching from the rear at a speed which would result in a collision within 3 seconds, the flashers turn on to warn the driver of the approaching vehicle to slow down. The specifications for the Altra Technologies Rear Guard System are as follows: 1. Horizontal beam width: +/- 15 degrees 2. Vertical beam width: +/- 20 degrees 3. Operating range: 0 to 300 feet Detailed Design Report 58 August 10, 2001 4. Maximum range error: 10% of reported range (2.0 feet minimum) 5. Maximum range rate: 100 mph The Rear Guard System mounts on 2 bars that extend up and to the rear from the top rear corners of the dump box on the snowplow. There is one flasher on the top of each bar. The intend is to place the flashers above the snow cloud that develops when plowing powdery snow. The sensor mounts on the right side bar below the right flasher. It faces straight to the rear horizontally, but is angled up at 10 degrees to allow for a dump box angle of up to 30 degrees. The only interface with the vehicle (or with the guidance system) is a control line which activates the Rear Guard System when pulled high (+12V). This allows the system to be inactive when the vehicle is stopped or in reverse. 3.1.3 Vehicle computer system The hardware platform used for all on-board vehicle driver assistive system computations is based on the PC 104 compact industrial computer standard. These computer components are essentially ruggedized single board computers which are hardened versions of standard PC’s. Both the ISA and PCI bus are available, but with unique connectors to PC 104 standard peripheral boards. All hardware components used in these computers are standard off-the-shelf items. Because of the computational load needed to compute vehicle state, complete database queries, and project graphics onto the HUD, two PC 104 computers are used for on-board computations in each vehicle. These two computers communicate via a dedicated local ethernet network. Each of these computers is equipped with a PC 104 to PCI adapter card; one computer uses a standard sound blaster compatible sound card, and the other uses a 3DFX Voodoo 3 graphics adapter. The PCI cards are used because at the present time, there exist no audio or video cards in the PC 104 format for which drivers for the QNX operating system are available. The on-board computers run the QNX Neutrino operating system. This is a hard real-time OS with deterministic scheduling capabilities. Neutrino is gaining market share, and as such, the availability of device drivers for many peripheral devices is improving. In the not to distant future, it is expected that device drivers for PC 104 format audio and video cards will become available, simplifying the mechanical stack up of the on-board vehicle computers. Table 4 summarizes the components which comprise the vehicle computers. The components were assembled into custom cases. The cases are aluminum, and the cases serve as a convection source for component cooling. The only active cooling device is the fan on the CPU. The face and inside of the custom computers are shown in pictures 22 – 24 on the following pages. The data flow into and out of the Vehicle computers is shown in Figure 22. Detailed Design Report 59 August 10, 2001 Component Vendor/Model Comments Single board computer, AMD k6-2 400MHz processor 16 channels A2D, D2A, 24 DIO Lines, 100KHz Main Board VersaLogic/VSB-6gp D2A, A2D, DIO board VersaLogic/VCM-DAS-1 Disk On Chip MSystems/DOC-048 48 MB Disk on Chip Cable Set VersaLogic DEV-006 All interface and power cables Computer RAM VersaLogic/MM3S-128 128 MB ram Video Card 3DFX Voodoo 3000 PCI Diamond Systems JMM-512V512 Connect Tech/Xtreme 8 port RS 232/rs485 serial card 50 Watt Switching power supply 8 serial ports with one interrupt Adapts 2.5” hard drive to PC 104 use Power Supply 8 Serial Port adapter Hard Drive Adapter Tri-M systems/DA104 Hard Drive Travelstar 20GN 10 GB hard drive, ATA/66 PCI – to –PC 104 Adapter Douglass electronics PCI to PC104 card adapter Audio Card CreativeLab PCI 128 OEM Soundblaster compatible Computer Case LMB Heeger/UC-976 Aluminum utility box Table 4. PC 104 Vehicle computer components Detailed Design Report 60 August 10, 2001 Picture 22: Front panel of Vehicle PC104 based vehicle computer. Each computer can control 10 serial ports; 2 RS 485 ports are dedicated on each computer. Aluminum housing acts as heat sink. Detailed Design Report 61 August 10, 2001 Audio Output Fuse Ethernet Power Picture 23: Back of PC 104 computer showing output for audible feeback amplifier, ethernet, fuse, and power cables. Detailed Design Report 62 August 10, 2001 PCI Card Picture 24: Inside of PC 104 vehicle computer. Vertical card is PCI card installed using PCI to PC104 adapter card. Detailed Design Report 63 August 10, 2001 Driver Interface Touch panel Processor Board, 128 MB memory, 400 MHz AMD-K6-2 Processor, onboard ethernet FWD Radar (3 Extreme Serial Card, 8 Serial ports controlled with one interrupt Side Radar 3DFX Voodoo Video Card VCM-DAS-1 A2D, D2A, and DIO Card Steering sensors 40 MB Disk on Chip HUD 10 GB 2.5” hard drive and associated card Brake sensors 50 Watt Power supply Auxiliary Displays Ethernet Hub vehDAQ Processor Board, 128 MB memory, 400 MHz AMD-K6-2 Processor, onboard ethernet GPS Recvr Extreme Serial Card, 8 Serial ports controlled with one interrupt RF Modem PCI Sound Blaster Audio Card IMU VCM-DAS-1 A2D, D2A, and DIO Card 3M sensor mag. sensor 40 MB Disk on Chip Audio Amplifier 50 Watt Power supply Seat Control Blue – RS 232 Red – RS 485 Plum – ethernet Green – VGA Turquoise – DIO Very green - analog Detailed Design Report Figure Figure 22. CFD. Data flow Datafor flow vehicle for vehiclePC PC104 104computers computers 64 August 10, 2001 3.1.4 Driver Interface Hardware Tree primary operational interfaces constitute the human-machine interface: the graphical user interface (GUI), the haptic user interface (HUI), and the audible user interface (AUI). The GUI is executed as a conformal, augmented display through a Head Up Display (HUD). The HUI is executed via a seat equipped with the ability to vibrate either side of the seat in the thigh and back areas. Finally, the AUI is executed via a computer based sound card, an audio power amplifier, and stereo speakers which provide directional tones (or earcons) to the driver through a pair of stereo speakers. Because the HUI and AUI are purchased from commercial sources, only a brief description of those components will be provided. HUD The HUD consists of three components; the combiner (and its associated mounting hardware), the projector (and its associated mounting hardware), and the computer software which drives the images provided by the projector. A simple schematic of the system is shown in Figure 23 below. Vehicle Roof Windshield Combiner Projector Driver Figure 23. Simplified HUD system layout. View looking at the left side of the vehicle. Detailed Design Report 65 August 10, 2001 The system works in the following way. The driver, during low visibility conditions, folds the combiner down to its operational position12 which is approximately 18 inches in front of the driver’s face. The driver then turns on the projector, which is located adjacent to the driver’s right ear. The projector then provides the images which are used by the driver to guide the vehicle during low visibility conditions. The driver sees through the combiner, and also sees reflections of the projector in the combiner. The reflective properties of the combiner are best illustrated by Picture 25 below. The potato head figure is reflected, but the background on the other side of the combiner is also visible. Note that the combiner does reduce the intensity of light passing through it. This is due to both the reflective coating on the concave side, and the anti-reflective coating on the convex side. Picture 25: Potato Head action figure illustrating both reflective and transmissive properties of the combiner. Note that the reflectivity is less than a fully reflective mirror. The combiner13 is partially reflective, partially transmissive optical device designed specifically for this purpose. The combiners used in the field operational test are made of 12 13 The mechanism which supports the combiner is discussed in the mechanical design section. Also commonly referred to in the optical world as a beam splitter. Detailed Design Report 66 August 10, 2001 chemically tempered optically ground glass,14 ground and coated to University of Minnesota specifications. Chemical tempering strengthens the optical substrate, and in the unlikely event the combiner is broken, allows the broken glass to come apart in smooth pieces. The coating on the concave surface is reflective; the reflective properties are designed such that all colors from the projector are reflected with nearly equal intensity. The level of reflectivity has been optimized for bright to zero ambient light conditions. For extremely bright conditions, (i.e., sunny days with new snowfall), a “contrast” enhancer is overlayed on the combiner frame. This contrast enhancer acts like a pair of sunglasses, and reduces the amount of light passing to the driver through the combiner. The enhancer is a piece of neutrally colored Plexiglas which is attached to the combiner frame with Velcro. An anti-reflective coating covers the convex side of the combiner. This anti-reflective coating improves contrast and eliminates glare. The effectiveness of the anti-reflective coating is best illustrated in the figure below. Note the glare in the windshield, and the lack of glare in the view through the combiner. Picture 26: HUD display under bright conditions. Note lack of glare through the combiner. The projector is based on a super bright LCD flat panel display, originally designed for police mobile data terminals. The particular combiner chosen for this system is the 10.4 inch diagonal Mobile VU display manufactured by Litton Systems. The combiner 14 Because of the small number of systems made for this FOT, economics dictate that the combiners be made of glass. In production quantities, the combiners would be made from an injection moulded and optically coated piece of acrylic or polycarbonate. These combiners would be much thinner, lighter, and less costly than the ground glass ones used in this project. Detailed Design Report 67 August 10, 2001 reflectivity specifications were derived based on the optical properties of the projector. The HUD is really a system, not an aggregation of components. The structural hardware which support both the projector and combiner are discussed in the following chapter. The Haptic User Interface consists of a seat which is capable of providing vibrational cues to the driver through the seat cushions. The vehicle seat is modified so that four motors are embedded in the seat cushion; two motors are placed on the left and right sides in the bottom support, and two motors are placed on the left sides in the seat back. These motors are equipped with an offset weight which provides for a dynamic imbalance. The intensity (and frequency) of the vibration is a function of the speed at which the motors rotate. For this project, the seats are custom built by Seat Comfort Systems, Bellflower, CA. The driver’s seat is removed from the vehicle, sent to Seat Comfort systems for modifications, and then returned to be re-installed in the vehicle. The seat controller provides a computer interface which allows for left seat support, left seat back, right seat support, and right seat back motors to be turned on and off individually. Vibration intensity is controlled by a single potentiometer so that all motors vibrate with equal intensity; individual zone intensity is not an option. Side Collision Avoidance Interface The visual side collision warning is a pair of Altra Technologies Mirror-mount LED Displays (MLD). The MLD is shown in Picture 27. It is a sealed unit intended to mount on the outside edge of the rear view mirrors on a truck. It mounts directly to the surface of the mirror using double sided adhesive foam tape. The dimensions of the MLD enclosure are 0.75 inch x 0.5 inch x 6.5 inch. The MLD consists of a vertical row of 6 sunlight LEDs. With the cable exiting the enclosure at the bottom the top LED is D1 and the bottom LED is D6. D1 is amber, D2 is red, D3 – D5 are amber, and D6 is red. LEDs are activated by pulling the return pin to ground. The interface to the MLD is a Deutsch DTM04-8PA connector. The side collision warning is activated if an object is within the side sensor field of view and if the turn signal is active. This warning is provided to the driver with the red LEDs. A side collision advisory is activated if a vehicle is located within the side sensor field of view. In this case, the advisory is provided via the amber LEDs. Detailed Design Report 68 August 10, 2001 Picture 27: LED based side collision warning interface Audible User Interface The hardware used to implement the Audible User interface consisted of a 35 watt stereo power amplifier and a pair of stereo speakers. The components are listed in table 5 below. Component Vendor/Model Comments Audio Amplifier Blaupunkt / MPA 160 35 Watts per channel Stereo Speakers Kenwood / KFC-x136 5.25” diameter speaker Table 5. Audible User Interface hardware The objective of the Audible interface is to provide to the driver a left or right directional cue or warning. To ensure that the cue is directional, the stereo speakers are mounted in the left and right driver’s doors. When a left warning is required, only the left channel is activated, and vice versa. The actual hardware is shown in Picture 28. Detailed Design Report 69 August 10, 2001 Picture 28: Audible User Interface Hardware 3.2 VEHICLE MECHANICAL AND ELECTRICAL DESIGN DETAILS In this section, the mechanical and electrical designs needed to support and power the equipment described previously are provided. The basic designs for all of the test vehicles are similar; for instance, GPS, radar, combiner, and projector mounts are identical for all six vehicles. The four snowplows share a similar hardware mounting configuration, but because four models from three different manufacturers (two Fords, one International, and one Sterling) are used for the test, some variations will arise. The ambulance and the State Patrol squad will be discussed in general terms. Hutchinson ambulance has provided an entire equipment cabinet for the installation of test hardware; all that is required for the installation of the equipment is the design and manufacture of some custom shelves. This will be a simple installation. The State Patrol Squad, however, will be a different animal. Because of the nature of the equipment carried by the Patrol (shovels, shotguns, surveillance equipment, etc.), the amount of space available for mounting the needed equipment in a squad is rather limited. To ensure that the installation of the equipment which runs the driver asssistive system and collects operational data doesn’t interfere with the day to day function of the patrol squad, a close working relationship with the company who outfits the State Patrol squads has been established.15 15 At the time this draft document was prepared, the vehicle to be used by the state patrol for this test had not been identified. All that can be said about the equipment installation at this time is that most of the equipment will be located in the truck. A number of options have been discussed, but nothing has been finalized yet. Detailed Design Report 70 August 10, 2001 Figure 24. GPS mount for snowplows. Hole at top has clearance for 5/8” x 11 TPI bolt. Bottom clamps to existing snowplow light bar. Mount is 24” hig to provide clear view to orbiting GPS satellites. The remainder of this section will follow a path parallel to that of the component path; first, mechanical hardware specific to the vehicle guidance system will be discussed, followed by radar, computers, and finally, the user interface hardware. Subsequent to that will be a discussion regarding electrical needs and designs needed to implement these systems. 3.2.1 Vehicle Guidance system Three components require support on the outside of the vehicle: the GPS antenna, the RF Modem antenna, and the magnetic sensor bar to detect proximity to the magnetic tape embedded in the roadway. These three components are described herein. GPS antenna mount The job of the GPS antenna is to receive the low power GPS signals from the GPS satellites in orbit about the earth. Because of the low power of the GPS satellites, a clear view of the satellites is necessary to ensure signal reception. This, in turn, requires that the GPS antenna be located high on the vehicle and not be blocked by any vehicle component. The GPS antenna mount designed for snowplows is shown in the Figure 24 below. The clamp at the bottom of the mount clamps to an existing light bar mounted on the roof of the truck cab. The hole at the top is drilled for clearance for a 5/8” x 11 TPI bolt. The GPS antenna has a 5/8” x 11 TPI insert at its bottom. The bolt acts to clamp the antenna to the mount. The GPS antenna for the ambulance will be mounted on a similar pedestal so that the antenna is located above the ambulance box. The GPS antenna for the State Patrol vehicle will be mounted above its lightbar. Detailed Design Report 71 August 10, 2001 The RF Modem antenna has two mounting options; a magnetic mount or a gutter/trunk lid mount. The mounting will be left to the discretion of the vehicle owner. Gutter or trunk mounts clamp to the vehicle rain gutter or trunk lid; magnetic mounts affix themselves to any ferrous material. Both options are standard off-the-shelf items with zero lead time. The vehicle owner will guide the selection process here. 3M Magnetic sensor mount Throughout the development process, there have been significant development improvements in the mounting hardware for the magnetic sensor. The mounting hardware will mount on the I-beam bumper similar to that shown in Figure 25 below. There are two variations of this mount to accommodate short and long bumper configurations. Figure 1shows the construction necessary for the long bumper. This design allows for a rugged mount to withstand the abusive environment but also allows flexibility in sensor positioning. This mount has been tested on the Highway 19 snowplow during the 1999 – 2000 winter. Snowplow Bumper Supports Sensor Bar Figure 25. Mounting system for magnetic sensor for typical snowplows Detailed Design Report 72 August 10, 2001 A very similar mount has been developed for the shorter I-beam bumper. This design can be used if a long bumper is not available. This design is shown in Picture 29 below. Picture 29: Short I beam magnetic sensor mount 3.2.2 Radar Mounts Forward Radar Mounts Three types of radar mounts are used for this project: grill mounts (specific to snowplows), forward mounts, and corner mounts. These are all illustrated in Figure 26 below. The grill mount, as its name implies, is mounted on the grill of the snowplow. This radar is aimed to look over the front blade when the blade is lowered to its snow moving position. It is aimed straight ahead. This mount allows one degree of freedom, rotation about the axis parallel to the vehicle longitudinal axis. The radar forward mount is also located on the vehicle longitudinal axis and it, too, looks straight ahead. It is typically mounted on a tubular structural member using the “U” bolts shown in the drawing. U bolt mounts again allow for one degree of freedom mount; rotation about an axis parallel to the vehicle lateral axis. For the roof mounted radar units, the tubular structural member is usually a Yakima bike rack. Finally, the third radar mount is the corner mount. It is similar to the forward mount, but it offers an additional degree of freedom: rotation about an axis parallel to the vehicle vertical axis. This additional degree of freedom allows the radar sensor to be installed so that it faces away from the vehicle longitudinal axis. The corner mount provides +/- ten Detailed Design Report 73 August 10, 2001 degrees of rotation about the vertical axis in addition to rotation about an axis parallel to the vehicle lateral axis. The radar mounts can easily be adapted to ambulance or squad car use. Hutchinson ambulance uses a tubular steel fixture to hold sirens and emergency lights in front of the grill. In Minnesota, the State Patrol uses a tubular steel “cow catcher” push bar in the front of their squad car grills. It is a straightforward task to adapt the radar mounts described above to these vehicles. Radar Corner Mount Grill Mount Radar Forward Mount Figure 26. Eaton Vorad Forward –looking radar mounts Detailed Design Report 74 August 10, 2001 Side Radar Mounts There are 2 mounting methods for the side sensors. When mounting on the vehicle body, or other panel, 3 holes (2 mounting holes and 1 cable hole) are drilled in the panel. This mounting method may be used on the ambulance and patrol car. When mounting on a truck step or framework a steel bracket is used to mount the sensor. This mounting method will be used on the snowplows. Rear Radar Mounts The Rear Guard System mounts on 2 bars that extend up and to the rear from the top rear corners of the dump box on the snowplow. There is one flasher on the top of each bar. The intend is to place the flashers above the snow cloud that develops when plowing powdery snow. The sensor mounts on the right side bar below the right flasher. It faces straight to the rear horizontally, but is angled up at 10 degrees to allow for a dump box angle of up to 30 degrees. 3.2.3 Driver Interface Mounting Hardware Only two mounts are required for the driver interface: the projector mount and the combiner mount. Both the projector and the combiner are attached to the roof structure of the vehicle. The projector mount is located over the driver’s right shoulder and is attached to the roof structure. The mount has three degrees of freedom: translation about the long axis of the main support channel, rotation about the vertical axis, and rotation about the lateral axis. Because the projector is reasonably massive, it is usually braced with a structural member which connects the floor to the roof. This avoids an “oil canning” effect for the roof caused by the vertical motion of the suspended projector mount. A typical projector mount assembly is shown in Figure 27. A more complicated driver interface mounting mechanical device is the combiner mount. First, the HUD has fold out of the way when it is not needed. Second, because of the variety of heights and sizes of the drivers who will use the HUD, it needs to be “aimed” so that proper alignment of the projected lines with those actually painted on the road are aligned. Third, it needs to be rigid to avoid vibration and the resultant image “jitter” associated with the vibration but at the same time be flexible and move out of the way should it be hit by any reason by a driver’s head. These goals have been met through the use of a four-bar mechanism and delrin washers at the linkage points. The function of the combiner mount is illustrated in Figure 28 and Pictures 30-31. Detailed Design Report 75 August 10, 2001 Figure 27. Top structure is easily modified to accommodate a wide variety of vehicle types. This structure is typically also supported by a structural member from the floor so that an “oil can” effect is avoided. The floor-to-roof structural member is custom designed for each vehicle type. Detailed Design Report 76 August 10, 2001 Figure 28. A. Design drawing of the combiner mount illustrating the available degrees of freedom. The rotational DOF shown by the top right arrow allows the combiner to be folded up; the rotational DOF on the left allows rotation for HUD aiming; the vertical translational DOF allows the height of the combiner to be adjusted to accommodate drivers of varying heights. Detailed Design Report 77 August 10, 2001 Delrin Washer Picture 30: Combiner in folded up position. Delrin washer is barely visible behind tension nut. Picture 31: Combiner in normal operating position (folded down) Detailed Design Report 78 August 10, 2001 3.2.4 General Mounting Hardware As has been indicated, considerable equipment is needed to operate the driver assistive system and the vehDAQ system described in the next chapter. For the snowplows, the equipment is located in the cab of the vehicle in custom designed boxes. This keeps exposure of sensitive equipment to the elements at a minimum. Because of the relatively large space offered in a snowplow cab, there is enough room to mount the equipment inside. As mentioned previously, Hutchinson ambulance has made a full size equipment cabinet available for the installation of the driver assistive system and vehDAQ data acquisition system. Installation in that situation will be very straightforward. The State Patrol squad is a different story, because most equipment will be located in the trunk. Precisely how this equipment will be located has yet to be determined. A layout of the equipment to be installed in the vehicles is provided in Figure 29 below. The boxes in which sensors and computers will be mounted are shown in Pictures 32-33. Figure 29. Plan layout typical of the equipment configuration for a snowplow Detailed Design Report 79 August 10, 2001 Picture 32: Snowplow equipment cabinet. Aluminum box replaces seat mount in Sterling cab. Other snowplows receive similar packages. Picture 33: Same box with protective covering Detailed Design Report 80 August 10, 2001 3.2.5 Electrical System Design Snowplows and state patrol squads, are typically delivered with alternators capable of a maximum current delivery of 120 to 130 amperes. Discussions with snowplow maintenance personnel and the Minnesota State patrol indicate that nearly all of this current is accounted for with existing vehicle systems16. The driver assistive and data acquisition package provided for these vehicles draw between 30 and 40 amperes. To provide sufficient electrical power, standard factory provided alternators are exchanged for high output alternators. All four snowplows are equipped with a Leece-Neville 220 amp alternator provided by EGR products in Canton, Ohio; that alternator is shown in Picture 34 below: Picture 34: 220 Amp Leece-Neville Alternator used for snowplow electrical power All of the equipment needed for the driver assistive system and the vehDAQ operates from a nominal 12 Volt source. To supply power to each component, a 12 volt power bus is located in the central equipment bay. The power bus consists of 16 power ports; each 16 Because of the life critical mission of the ambulance, those vehicles are supplied with either additional alternators or specially designed high output alternators. Therefore, no additional power is needed for the ambulance to support this field operational test. Detailed Design Report 81 August 10, 2001 port is terminated with an XLR connector typically used for sound reinforcement equipment as found at rock and roll shows. The XLR is rugged, waterproof, and uses a positive engagement mechanism which prevents a connection from coming loose. An example of an XLR connector is provided in Picture 35 below. The complete bus is shown in Picture 36. Picture 35: A female XLR connector. Note weatherproof gasket and positive engagement mechanism. Picture 36: Power bus faceplate. Each row is independently fused. XLR ports are used to provide power to all 12 volt equipment used for vehDAQ and driver assistive system. In addition to 12 volt power, a source of 120 VAC is provided. This voltage source is made available as a development and debug convenience for laptop computers and other analysis equipment. Two standard GFI outlets are provided on for the 120 Volt power. The 12 VDC source is converted to 120 VAC with an Exeltech XP-1100 inverter, capable of supplying 1100 steady state watts, and 2200 peak watts. The inverter is shown in Picture 37. Detailed Design Report 82 August 10, 2001 Picture 37: ExelTech 1100 Watt Inverter Detailed Design Report 83 August 10, 2001 Detailed Design Report 84 August 10, 2001 4.0 DRIVER ASSISTIVE SYSTEM (DAS) INTERFACE DESIGN 4.1 INTRODUCTION AND BACKGROUND This section of the report describes two driving simulator experiments and a field test that were undertaken to explore relevant human factors aspects of the Driver Assistive System (DAS) Interface. The overall objective of these studies was to produce design recommendations for this interface. After briefly describing the display design guidelines, the experiments conducted in a driving simulator and the field study at a test track are described. The information gained from the driving simulator experiments and the field study is channeled into design recommendations for the DAS Interface at the end of the section. 4.2 DISPLAY DESIGN GUIDELINES The design guidelines described here were derived, in part, from general guidelines suggested by Norman (1998) and from more specific guidelines developed by European and U.S human factors experts involved in the design of intelligent driver support systems (Michon, Piersma, Smiley, Verwey, & Webster, 1990; Ng & Barfield, 1998). In order to efficiently design displays that are to be used in driving applications, it is necessary to take into account the dynamic nature of the environment in which the displays are employed. The driving environment continually changes as the driver proceeds along a road and encounters and interacts with other vehicles, intersections, and signs. For this particular application, as a specialty vehicle moves through this dynamic environment, the information provided by the DAS will continuously be updated. The design guidelines we used in considering the DAS Interface involved: (1) The DAS should improve safety. (2) The full capabilities and of the systems supplying information to the DAS should be utilized. (3) The limitations of the systems supplying information to the DAS should be considered and respected. (4) Information should be provided to a driver intuitively. (5) Conformality with real world objects and their representations will ensure effective information delivery. (6) Latency associated with information transfer should be minimized. (7) Existing standards will be used where possible. Detailed Design Report 85 August 10, 2001 4.3 FUNCTIONS OF THE DAS INTERFACE The DAS Interface provides the specialty vehicle operator with the following: (1) Lane markings for lateral and longitudinal guidance. (2) Lane departure warnings for lateral guidance. (3) The presence and location of objects which coud cause a collision. (4) Collision avoidance warnings. The lane markings and lane departure warnings are derived from inputs provided by the DGPS and the geospatial database. The collision avoidance warnings are provided by a forward-looking radar system that detects and locates obstacles ahead of the specialty vehicle. When the DAS is deployed, the operators will be able to use this information to combat the effects of poor visibility if they are confronted by blowing or drifting snow, whiteout conditions, or fog, night, etc. The operators will be able to use the lane markings and lane departure warnings to stay in their lane and the information about the presence and location of potential obstacles to avoid collisions. The positional and collision avoidance information is presented via the DAS Interface. This interface consists of three displays— (1) Head-Up Display (HUD). Visual information is provided via the HUD which consists of a combiner mounted between the driver and the windshield, and a projector positioned just to the right of the operator (2) Auditory Display. Auditory information is provided via four speakers mounted in the four corners of the vehicle’s cab.17 (3) Active Seat. Vibrators installed in the left and right sides of the driver's seat bottom provide directional haptic information. 4.3.1 Lane Markings The HUD provides continuous positional information by showing the edge- and centerline markings of the road ahead of the operator. When viewed by the operator, the edge- and centerlines of the image on the combiner should overlay the actual edge- line and centerline on the road ahead, so that in conditions of very poor visibility, the driver will be able to drive by using the projected edge- and centerline markings instead of the actual lane markings. 17 For the Human Factors studies, four speakers were used both in the simulator and in the experimental vehicle. However, due to operational considerations (rear speakers being kicked out of the doors of a state patrol squad) and size constraints (two-passenger snowplow and ambulance cabs), only two speakers will be used for field operational test vehicles. Detailed Design Report 86 August 10, 2001 4.3.2 Lane Departure Warnings Lane departure warnings can be given to the operator as soon as one of the wheels of the operator’s vehicle crosses a limit set by a driver18. These warnings can be presented via the HUD, the auditory display, and/or the active seat. The experiments described in this section of the report investigated alternate ways of presenting these warnings. 4.3.3 Collision Avoidance Warnings The specialty vehicle operator is particularly concerned with vehicles ahead that are stationary or traveling more slowly. Warning the operator of the presence of such potential obstacles will allow him or her to take appropriate action, such as slowing down and/or navigating around the vehicle ahead. The DAS includes a forward-looking radar that can detect and locates objects that potentially could become obstacles. This includes moving and stationary objects that are in the lane ahead, in adjacent lanes, in the entrance of intersecting roads, and on the hard shoulder within the plus or minus 15 deg sweep of the radar beam. The human factors issues associated with delivering collision avoidance warnings relate to the symbology used to deliver the warning, and the timing of that warning. These two issues are dealt with in the subsections below Collision Warning Symbology It should be noted that the DAS radar only detects and locates objects; it cannot identify objects, or determine their size. So, in order to maximize safety, when the radar does detect an object in the road ahead it is recommended that the warning symbol presented to the operator on the HUD should be a rectangular outline that is the width of the widest vehicle that a driver is likely to encounter—i.e., a truck. This will ensure that the driver is prepared for the worst-case possibility. When the DAS radar detects an object ahead, in addition to creating a warning symbol that when it is presented on the DAS Interface is the width of a truck, it is also recommended that the symbol should also approximate the shape of a truck—in particular that it should have a flat top. Giving the symbol a flat top will make it conform with the standard way in which trucks are depicted on U.S. highway signs—see for example the trucks depicted in Figure 30 below. The figures are taken from the FHWA Manual on Uniform Control Devices (MUTCD, 2000)—and, as Norman (1998) points out, wherever possible, it is a sensible design procedure to conform to existing standards. In addition, the width of the rectangle should be proportional to the width of the crosssection of a truck at the distance that the detected object is from the driver. Because the width of an object will increase as the operator approaches it, the rectangular symbol 18 For snowplow operations, operators typically plow a few feet out of lane to ensure complete snow removal. In these situations, the operator prefers to choose his or her "out of lane" limits. Detailed Design Report 87 August 10, 2001 should dynamically increase in width as well, in order to maintain the proportional relationship between the width of the symbol and the width of the truck it represents. Figure 30: The U.S. Standard National Network Signs (R14-4, R14-5) (from the FHWA Manual on Uniform Control Devices, MUTCD, 2000). Dynamically increasing the size of the warning symbol as the operator approaches the object—in addition to presenting a warning—will give the added benefit of providing a depth cue for the driver. Finally, the rectangular warning symbol should only be outlined. The reason for this is so that when the operator gets close enough to it, he or she will be able to see the details of the object within the frame of the outline, and thus be able to determine what the object actually is—i.e., whether it is a truck or whether it is a smaller vehicle. As well as allowing the operator to look through it, the outline’s frame should be sufficiently wide (perhaps 10 % of the symbol width) that it will provide a clearly perceivable warning. One of the guiding design principles is to make the display as conformal as possible. An open rectangle helps to ensure that drivers will not be hindered from identifying what the object is. Collision Warning Timing With regard to the timing of the collision avoidance warnings, Färber’s (1991) argument about designing them in such a way as to minimize false alarms is of relevance. He argues that collision avoidance warnings should be delayed so that they err in the direction of high validity (even though, as a result, there may be occasions when the situation is relatively dangerous and the system is not activated). Then the driver will know that, when the collision alarm system is activated, it is activated with good reason, and that he or she should respond to it. Färber suggests that having warnings that are delayed is preferable to having a collision avoidance system that is activated early in all situations that pose some danger, because the latter system would inevitably have a relatively high false alarm rate, so that the driver could become annoyed and might deactivate it (just as commercial airline pilots deactivated their initial Ground Proximity Warning Systems because of their extremely high false alarm rate).19 19 Such system reliability is achieved with the advanced radar processing algorithm which uses the geospatial database to verify the validity of radar returns. Detailed Design Report 88 August 10, 2001 Van der Horst’s (1991) study is also of relevance. He had subjects drive an instrumented car on a closed track toward a stationary object at speeds of 18.6 mph (30 km/h), 31.1 mph (50 km/h), and 43.5 mph (70 km/h). When his subjects were instructed to start normal breaking at the last possible moment, van der Horst found that the time-tocollision was less than 3 s for all three speeds. This suggests that 3 s is the lower limit for the criterion we should adopt for the DAS collision avoidance warning. When van der Horst’s subjects were instructed to start hard braking at the last possible moment, he found that the time-to-collision dropped to less than 2 s for all three approach speeds— this suggests that a 2 s onset of a warning might be a little too risky for the DAS warnings. In a later study, Maretzke & Jacob (1992) indicated that a longer time-to-collision, such as 5 s might be more appropriate. We decided to use this interval as an advisory time in a sequence where a driver approaching a stationary object might be given an advisory at 5 s and a warning at 3 s.20 Because a vehicle traveling at 45 mph (72.5 km/h) travels a distance of 198 ft (60.4 m) in 3 s and 330 ft (100.7 m) in 5 s, a DAS radar range of 350 ft (106.8 m) will probably be sufficient for snowplow operators, who do not have high-speed requirements. However, the range of 350 ft (106.8 m) provided by the DAS radar is less likely to be adequate for the other two groups of specialty drivers—ambulance drivers and highway patrol drivers—since a vehicle traveling at 75 mph (120.8 km/h) covers 330 ft (100.7 m) in 3 s and 550 ft (167.8 m) in 5 s. 4.4 SIMULATION EXPERIMENT 1: THE EFFECTS OF VARYING VISUAL LANE DEPARTURE AND COLLISION AVOIDANCE WARNINGS 4.4.1 Introduction The purpose of this driving simulation experiment was to explore ways of presenting lane departure and collision avoidance information visually on the HUD. It did not involve the other two displays that are part of the DAS Interface (i.e., the auditory display and the active seat). The effectiveness of various formats for visually presenting lane departure and collision warnings was investigated. There were five visual lane departure warning conditions. One was a no-warning control condition. In the other four conditions, as soon as the driver’s front tire touched a lane marker, whether to the left or the right of the driver, a lane departure warning was given. The warning involved a change to the projected lane marker and/or to the area into which 20 We did not plan to test the situation where the driver approaches a slowly moving vehicle in the three studies reported here--we will consider this situation in analytic studies currently being planned. Detailed Design Report 89 August 10, 2001 the driver was moving. The lane departure warning conditions are listed below. Examples of each condition are shown in Appendix B, Figures IV-B1 to IV-B5. (1) No-warning control condition. (2) Red line warning condition, in which the line being crossed changes color from white to red on the HUD. (3) Double line warning condition, in which the line being crossed changes from a single white line to a double white line on the HUD. (4) White area warning condition, in which the area on the HUD immediately adjacent to the line being crossed becomes opaque and white. (5) Red area warning condition, in which the area on the HUD immediately adjacent to the line being crossed becomes opaque and red. Picture 38: No-warning control condition Detailed Design Report 90 August 10, 2001 Picture 39: Red line warning condition, in which the line being crossed changes color from white to red on the HUD Picture 40: Double line warning condition, in which the line being crossed changes from a single white line to a double white line on the HUD. (Shown in close-up) Detailed Design Report 91 August 10, 2001 Picture 41: White area warning condition, in which the area on the HUD immediately adjacent to the line being crossed becomes opaque and white. Picture 42: Red area warning condition, in which the area on the HUD immediately adjacent to the line being crossed becomes opaque and red. Detailed Design Report 92 August 10, 2001 In addition to the lane departure warnings, the HUD can present continuously updated information about the presence of objects ahead of the driver’s vehicle. When the vehicle and an object ahead are on a collision course (because the object is in the path of the driver’s vehicle and is either stationary or has a slower velocity than the driver’s vehicle) two visual indications of a possible collision may be given. The first, a visual collision advisory, may be given if the time to collision falls below 5 s and the second, a visual collision warning, may be given if the time to collision falls below 3 s. In this experiment, three visual collision avoidance warning conditions were tested. Collision avoidance advisories and warnings were indicated by changes in color to the white outline that indicates the presence of an object ahead. The conditions were (1) No-warning control condition. (2) An advisory & warning condition, in which an advisory was given when the time to collision reached or fell below 5 s, and a warning was given when the time to collision reached or fell below 3 s. The advisory was indicated by a white rectangular outline of the object ahead changing to yellow and the warning was indicated by the yellow rectangular outline changing to red. (3) Warning only condition, in which a warning was given when the time to collision reached or fell below 3 s. The warning was denoted by a red rectangular outline of the object. 4.4.2 Method Participants Participants responded to flyers posted around the University of Minnesota’s Twin Cities campus. There were 15 participants between 18 and 65 years of age (10 males and 5 females). Each had a valid driver’s license at the time of the experiment. Participants were reimbursed $10 for their time. Apparatus Participants piloted the wrap-around driving simulator (WAS) at the University of Minnesota’s Human Factors Research Laboratory. The WAS is a fixed base driving simulator, consisting of a full-body 1990 Acura Integra RS enclosed in a spherical wood and steel dome measuring 12.5 ft (3.81 m) high at its apex with a 15.5 ft (4.73 m) internal diameter. The car was fitted with sensors to detect accelerator, brake, and steering inputs, providing a real-time interface for the driver. Force feedback was applied to the steering wheel through the steering column with a torque motor. The speedometer was functional, and powered by a small motor that was controlled by the main simulator computer. The virtual environment through which the participants drove was generated with an SGI Onyx computer (Reality Engine 2). The programming was done using MultiGen- Detailed Design Report 93 August 10, 2001 Paradigm Vega and SGI Performer APIs. The main simulator computer was a PC running Linux, which processed all vehicular sensors and controllers. The vehicular hardware interfaced the main simulator computer by means of a National Instruments AT-MIO-16E-10 data card. Information from this computer was transmitted to and from the Onyx via TCP/IP. The Onyx calculated the vehicle dynamics and generated the visual scenario. Three Proxima 9250+ projectors, operating at a resolution of 640x480 (a resolution that is lower than the capacity of the projectors, but stems from the constraints of the SGI Onyx computer), were used to create the visual scene inside the simulator. The virtual roadway was projected onto a curved seamless 24 ft (7.32 m) x 8 ft (2.44 m) screen positioned in front of the car. Together, the projectors provided a 156 deg forward view. Each projector received a discrete portion of the driving environment from the Onyx. The experimenter controlled the simulator from a remote station outside the dome. The experimenter communicated with the participant by means of a microphone (Audio Technica) relayed through a portable sound system (Sony #CFS-720, Tokyo) to a speaker placed in the back seat of the Acura. This sound system also had a cassette deck, which was used to deliver the instructions to the participant. Video information was relayed from inside the Acura to a television monitor at the experimenter’s station (JVC, Tokyo) by means of a miniature vehDAQ camera system. Miniature cameras were positioned on the rear-view mirror and on the passenger’s side B-pillar to record the driver’s face and arm movements. A separate low-light camera was mounted under the steering column and was used to record the position of the driver’s feet on the accelerator and brake. The signal from the center projector was captured using a scan converter and then input to the vehDAQ system. The image on the monitor was divided into quadrants, each one displaying one of the images mentioned above, as well as the current speed of the vehicle in the simulator and the elapsed time for the trial. The speed and time information was provided by the main simulator computer, which output the information to a titler device that overlaid the data onto the vehDAQ images. The experimental session was videotaped. A lavaliere microphone was mounted on the rear-view mirror to extract audio from inside the dome to the experimenter’s station. This audio signal was sent through the VCR and television as well. Engine/road noise was generated by the Onyx and fed through a Cerwin-Vega satellite and subwoofer system, mounted in the trunk of the vehicle, and two Aura bass shakers mounted under the front seats. A separate stereo receiver (Sony #STR-D365, Tokyo.) powered this supplemental system. Visual navigational information inside the car was conveyed via a Head-Up Display (HUD) developed by the University of Minnesota Intelligent Vehicles Lab. The HUD software was run on a Micron Pentium PC which received lane-position information from the main simulator computer. Information from the HUD was simultaneously overlaid onto the virtual environment. Images on the HUD were projected from a 10.4 in (266.7 mm) Diagonal Mobile VU display (Litton Systems #46830-1, San Diego) mounted to the Detailed Design Report 94 August 10, 2001 right of the driver’s headrest. The display projected the image onto the image combiner, a vertically and horizontally planar convex plate of glass mounted to the ceiling of the car directly in front of the driver’s head (the distance from the head ranged from 7 in (179.5 mm) to 12 in (307.7 mm) dependent upon where the seat was positioned). The combiner measured 11.5 in (294.9 mm) wide and 6 in (153.8 mm) high and was framed with clear 0.75 in (19.2 mm) polycarbonate. The driver looked through the combiner (which reflected the projected image to the driver) to the screen in front of the car onto which the virtual environment was projected. The position of the combiner was adjustable to accommodate drivers of different heights. Picture 43 below shows a lab research assistant driving with the DAS in the driving simulator. Picture 43: Driving with the DAS in the driving simulator 4.4.3 Experimental Design In this first experiment, two primary variables were investigated: (1) Lane departure warnings. (2) Collision avoidance warnings. Detailed Design Report 95 August 10, 2001 To investigate these two conditions, two mutually exclusive factorial designs were employed within the same experimental scenario. The experimental design was keyed first to the investigation of lane departure warnings, because this was the primary variable with the most levels; then, folded within it was the design used to investigate the collision avoidance warnings. In addition, three other variables (road width and the type and placement of stationary variable) were also folded into the design. However, their effects were not analyzed— they were included in the experiment only to provoke actions that would enable us to test the primary variables. Lane Departure Warnings A conventional factorial design was used to investigate the effects of varying the lane departure warning. It was varied as a between-subjects factor. The five lane departure conditions were: (1) No-warning control; (2) Red line warning; (3) Double line warning; (4) White area warning; (5) Red area warning. Each of the five lane departure warnings was presented to 15 subjects in five separate trials. For the purposes of randomization the subjects were assigned to three groups of five, with a different Latin square used to counterbalance the order for each group. Not only did this counterbalancing procedure minimize practice effects across subjects by presenting each condition an equal number of times in each trial, but also it limited (as far as is possible with 5x5 Latin squares) sequential carry-over effects from one warning condition to another. Collision Avoidance Warnings A conventional factorial design was also used to investigate the effects of varying the collision avoidance warnings. There were three collision avoidance conditions—(1) Nowarning control; (2) Advisory & warning; (3) Warning only. Because this variable had three levels and because its testing was folded within the design of the five-factor investigation of lane departure conditions, it was varied as a partially between-subjects and partially within-subjects factor. In this experiment, there were a total of 75 trials, with each of the 15 subjects taking part in five trials. The collision avoidance warnings were presented in a counterbalanced order by using all three possible 3x3 Latin squares in sequential order eight times (with one used nine times), and by folding the items remaining from the order for one subject into the order for the next. The result was that, for each subject, two of the three collision avoidance warnings were presented twice, while the third was presented once in the five trials that each received. Secondary Variables In addition to the two primary variables, three other variables—road width and the type and placement of stationary variable—were manipulated within the design of this Detailed Design Report 96 August 10, 2001 experiment to produce six unique routes. Although these three secondary variables were folded into the design, their effects were not analyzed—they were included in the experiment in order to provoke reactions from the subjects that would make it possible to test the primary variables. There were three road widths: 9 ft (2.75 m), 10 ft (3.05 m), and 11 ft (3.36 m). Each of these widths was used twice, to give the six routes. Two stationary vehicles—one half in lane, the other completely out of lane but touching the edge-line lane marker—were placed in each of the six routes. To determine how far from the start of the routes the stationary vehicles would be located, first the six routes were divided into five 1.24-mi (2-km) segments. No stationary vehicles were located in the first segment of any route. The segments of the routes in which the vehicles were located and the distance from the start of the segment were randomized. Table F1 (located in Appendix F) shows the route numbers, the lane widths, the segment in which the stationary vehicles were located, how far from the start of the segment that the vehicles were located, and whether the vehicles were half in or just out of lane. Overall Design The way in which the different design elements were combined is presented in Appendix F, Table F2. For each of the 15 subjects who participated in the experiment the table presents the route number, the lane departure warning condition, and the collision avoidance warning condition they experienced in each trial. 4.4.4 Procedure Participants drove through a virtual environment consisting of a two-lane highway composed of five 1.24 mi (2 km) segments. Between each segment was a cross-street 40 ft (12.2 m) wide including roadway and shoulders. For a given trial, participants were asked to drive on a highway with 9 ft (2.75 m), 10 ft (3.05 m), or 11 ft (3.36 m) lane widths. The presentation of lane widths was counterbalanced across collision avoidance warnings and lane departure warnings. Participants were not informed that lane widths would be variable. Vehicles appearing on the road were three-dimensional models with photos of actual passenger cars and sport utility vehicles texture-mapped onto the model. These vehicles were positioned at various intervals along the course. The lane position of the vehicles was varied as well with some placed alongside the road and some positioned partially within the driver’s lane. All vehicles were stationary. Detailed Design Report 97 August 10, 2001 Visibility for all experimental trials was controlled at 98.4 ft (30 m) by imposing a dense, virtual fog.21 In the HUD, solid white lines were projected onto the combiner to represent the fog line of the virtual road and a dashed white line was used to represent the centerline. Depending on the trial, participants were given one of four lane departure warnings or no lane departure warning. Counterbalanced across these warnings was one of three collision avoidance warnings, or no collision avoidance warning (see Appendix F, Table F2). If the participant steered the vehicle across either the fog or the center line in a condition where lane departure warnings were activated, the lane marker they had violated would either turn into a double line, change color to red, or the area outside the lane marker would be shaded either white or red. In trials with a collision avoidance warning, when the participant came within 350 ft (106.8 m) of a vehicular obstruction, a white rectangular outline of the offending vehicle appeared in the HUD. As the participant continued traveling toward the vehicle, the rectangle either stayed white, or changed to red with or without first passing through yellow. In conditions when the outline changed color, the yellow shift occurred approximately 5 s prior to the time of contact (estimated by the computer by factoring in the participant’s current speed). Similarly, the outline changed to red at a distance of 3 s. When the driver passed the offending vehicle, the outline disappeared. After signing the consent form, participants were taken to the driving simulator and given a training session to familiarize them with it. For the training trial, the combiner was not 21 Hawkins (1988) reports the results of several roadside observational studies investigating the relationship between low visibility and traffic speed. The poorest visibility range was 163.9 ft (50 m) to 491.8 ft (150 m)—which he defines as “thick fog.” In this experiment, it was desirable to have a very low visibility condition so that if the drivers were unaided (i.e., did not have a HUD) they would probably refuse to drive; To achieve this result a level of 98.4 ft (30 m), which is even worse than “thick fog, was chosen.” In experiment 2, there were snow covered roads as well as fog. When 98.4 ft (30 m) fog with the snow cover was used it was evident that when driving it was not possible to distinguish the ground from the fog. The fog level was decreased until the fog plus snow condition was of similar difficulty to the 98.4 ft (30 m) fog in experiment 1. In the two driving simulator experiments the virtual fog was modeled (Neider, Davis, & Woo; 1993) in the following way: The dense ground fog required for this experiment is obtained by using Neider et al’s equation “GL_EXP” which is as follows— f =e −( − ln( 0.001) z) v where v is the desired visibility [98.4 ft (30 m) in experiment 1, 295.1 ft (90 m) in experiment 2] z is the apparent distance between the driver in the simulator vehicle and a particular object in the scenario at a particular moment in time. f is the fog factor used in the simulation. The fog factor range is [0,1]. It is used as a weighting factor. If the fog factor is zero, then the object is given its original color. If the fog factor is one, then the object is given the fog color. And, if the fog factor has a value between zero and unity, then the object color and fog color are mixed proportionately—for example, if f = 0.7, the object would be 70% fog colored and 30% self colored. Detailed Design Report 98 August 10, 2001 installed in the simulator and the projector, though on, did not project an image. The training environment consisted of a straight, one-way, three-lane highway with no additional vehicles on the road. During the training session the experimenter stood to the left of the car behind the driver’s field-of-view. (For the actual experiment, the experimenter communicated with the participant from outside the dome using the system described above.) Participants were instructed by the experimenter to accelerate to 65 mph (104.7 km/h) and perform several lane changes and speed related maneuvers. After the requisite maneuvers were completed, participants were allowed to drive without experimenter direction until they felt comfortable using the simulator, at which point the experimenter installed the combiner for the HUD. After installing the combiner in the cabin, the experimenter presented another scenario that was used to help the driver adjust the HUD to a position appropriate for him or her. This scenario consisted of a stationary two-way stretch of road. Visibility was limited by fog but visibility was increased to 2,459 ft (750 m) so that the centerline and fog line could be seen. Subjects were instructed to move the combiner until, from their perspective, the HUD's roadway markings were aligned with the lines on the road. Following the combiner adjustment, the cassette containing the main instructions for the experiment was played for the participant. The instructions provided participants with background information necessary for them to take part in the experiment. A transcript of these instructions is included in Appendix F. The experimenter then answered any questions the participant had. When the participant stated an understanding of the information, the experimenter played the instructions for the first experimental condition. When the participant indicated he or she understood the instructions, the experimenter started the first experimental condition. Participants each ran in a total of five experimental conditions, with the experimenter presenting instructions and clarifying questions between each. These instructions are included in the Appendix. After the fifth condition, the participant was taken to another area where they were informally interviewed and debriefed by the experimenter. 4.4.5 Results Lane Departure Warnings In normal driving, lane departures occur infrequently. In order to induce more lane departures than normal, three lane widths, all less than the U.S. standard lane width of 12 ft, were used in this experiment. The widths were 9 ft (2.75 m), 10 ft (3.05 m), and 11 ft (3.36 m), and they were varied, in a counterbalanced fashion, from trial to trial. The participants were not informed what the lane width was in any trial, nor were they told that they varied from trial to trial. The time between the beginning of each lane departure and the moment at which the vehicle had completely returned to the lane was measured. The lane departure duration was used as a measure of the effectiveness of the lane departure warning conditions. Detailed Design Report 99 August 10, 2001 The first step in analyzing the lane departure duration data was to test for normalcy. As Winer, Brown, and Michels (1991) point out, normal distributions are particularly useful in parametric inferential statistics both because they provide a reasonable model for a very wide range of biological, sociological and behavioral phenomena, and because they are mathematically tractable. Since the pioneering work by Fisher (1935), the use of parametric inferential statistics has become pervasive in the biological, sociological and behavioral sciences and an immense literature has been produced describing numerous testing variants. Parametric statistics were once thought to require that the variance due to experimental error in the treatment variances should be homogeneous. However, since the work of Box (1954a, 1954b) parametric statistics are recognized to be robust with regard to heterogeneity of variance. However, they remain sensitive to skew or kurtosis. As Ferguson (1959) points out, skew (which is a function of the second and third moments of a distribution) measures asymmetry, while kurtosis (which is a function of the second and fourth moments) measures the extent to which a distribution is more or less peaked than a normal distribution. The closer the measures of both skew ( g1 ) and kurtosis ( g 2 ) are to zero, the more the distribution approximates the normal distribution. Figure 31 shows the overall distribution of the lane departure duration data that were obtained in this driving simulator study. When these data were tested for skewness and kurtosis, the values of g1 and g 2 were found to be— g1 =7.936 g 2 =82.773 Duration of Departure 140 120 100 80 60 Frequency 40 Std. Dev = 2.88 20 Mean = 1.7 N = 276.00 0 0.0 4.0 2.0 8.0 6.0 12.0 16.0 20.0 24.0 28.0 32.0 36.0 10.0 14.0 18.0 22.0 26.0 30.0 34.0 Duration of Departure Figure 31: Overall distribution of lane departure durations for first driving simulation study Detailed Design Report 100 August 10, 2001 As can been seen in Figure 31, the data are positively skewed and the distribution has a very long tail. It is appropriate for data of this form to be normalized by using a logarithmic transformation. After a logarithmic transformation was applied, the transformed data were distributed as shown in Figure 32. When the transformed data were tested for skewness and kurtosis, the g1 and g 2 values were greatly reduced. They were: g1 =0.444 g 2 =1.912 Log of Duration 60 50 40 30 Frequency 20 Std. Dev = .36 10 Mean = .05 N = 276.00 0 50 1. 25 1. 00 1. 5 .7 0 .5 5 .2 5 0 00 0. -.2 -.5 5 0 .0 -.7 -1 Log of Duration Figure 32: Overall distribution of logarithmically transformed lane departure durations for first driving simulation study. The next step was to conduct a 2x2 analysis of variance (ANOVA) on the logarithmically transformed data. The results of the ANOVA are shown in Table 6. Detailed Design Report 101 August 10, 2001 Table 6: Summary of 2x2 ANOVA conducted on the logarithmically transformed lane departure durations for first driving simulation study. (df = degrees of freedom and Subj… = Subjects) Table 6 shows that while both main effects (i.e., between departure warnings and between subjects) were statistically significant, the interaction between them was not. It is not uncommon to find a significant between-subjects effect. The comparison that is of most interest is that between lane departure warnings was significant at the p=0.018 level. To further elucidate the effect of the different lane departure warning conditions, post hoc testing was conducted using the Tukey Compromise (SPSS, 2000). The resultant pattern of significance is shown in Table 7. Condition No Warning Red Area Red Line Double Line White Area No Warning ---- Red Area 0.05 ---- Red Line 0.05 ns ---- Double Line 0.05 ns ns ---- White Area 0.05 ns ns ns ---- Table 7: Result of post hoc test showing the effect of lane departure warnings in the first driving simulation study (p<0.05). Table 7 reveals that the lane departure durations were statistically significantly shorter when the participants were given a warning than when they were not given a warning, but that there was no difference in the mean durations of the four lane departure warning conditions (the mean departure duration with no warning was 1.76 s, while it was half that for the four warning conditions—0.93 s for the red area; 0.92 s for red line; 0.87 s for the double line; 0.85 s for the white area 22). The mean durations for the four lane departure warning conditions are essentially the same; there is no statistically significant difference between them. 22 Each mean value given here is the antilog of the mean of the logarithmically transformed lane departure durations, for the particular lane departure warning condition. Detailed Design Report 102 August 10, 2001 Collision Avoidance Warnings In order to determine the effectiveness of the three collision avoidance warning conditions, we determined whether each participant reacted to the presence of the stationary vehicles that he or she encountered on each trial. We examined the driving performance of the participants in the 350 ft (106.8 m) section of the route before each target (i.e., the region within which the DAS radar could detect the stationary vehicle). We considered two possible reactions—reductions in speed, and steering maneuvers. Examination of the data showed that there were very few reactions that involved reductions in speed; so, this did not prove to be a useful measure. However, of 150 encounters (15 participants x 5 trials x 2 stationary vehicles) the participants responded with a steering maneuver on 119 occasions. The time and distance at which the steering maneuver took place were analyzed. However, there was no statistically significant collision warning main effect. While there were no differences in the steering response that occurred, there were differences in the number of responses that occurred. In the no collision warning condition, the participant gave a steering response in 68% of the encounters. In contrast, in the two collision warning conditions there was an increase in the number of steering responses to 84%, for the advisory and warning condition, and 82% for the warning only condition. 4.4.6 Discussion Lane Departure Warnings All four lane departure warnings showed similar reductions in lane departure times, when compared to the no-warning condition. On average, the participants returned to their lane twice as fast with the warnings as they did without the warning. However, because the four warnings were statistically inseparable, we had to use other criteria to select one for further testing. Two of the warnings obscured the driver's view of the road and another could be confused with universally used lane marking conventions. Both warnings that involved a large solid area (one red, the other white) would obscure any objects—including other vehicles—that happened to be in those areas. Because they would obscure and because neither area warning produced shorter lane departure times than the other two lane departure warning conditions, they were not selected. Moreover, the double-white line warning could be confused with the doubleyellow line warning that is used on roadways as the U.S. national standard for indicating that drivers should not cross the line into the opposing lane (MUTCD, 2000). Because this could potentially result in confusion and because it did not produce shorter lane departure times it was not selected either. Detailed Design Report 103 August 10, 2001 The fourth lane departure warning was the red line. It is both effective, visually salient, and consistent with the way red is used by the MUTCD (2000). For these reasons we selected it for further study. Collision Avoidance Warnings Both of the collision warnings resulted in a higher percentage of steering responses to stationary vehicles than not having a warning. However, based only on the steering response data there was no way to decide which of the two warnings would warn drivers most effectively. Post-experimental interview data indicated, like the steering response data, that the red warning was effective. However, most participants reported that they did not notice the change from white to yellow in the advisory and warning condition. When the mean speed at which the participants drove for each trial was examined it was found that the mean speed increased systematically from 37.7 mph (60.7 km/h) [sd = 11.4 mph (18.4 km/h)] in the first trial up to 47.7 mph (76.8 km/h) [sd = 15.2 mph (24.5 km/h)] in Trial 5. A vehicle traveling at 47.72 mph (76.84 km/h) will cover 350 ft (106.8 m) in 5 s. Because, the DAS radar is only activated when the driver is 350 ft (106.8) away from an object, a participant driving with the yellow-advisory-and-red-warning collision warning who is also traveling at a speed equal to or faster than 47.72 mph (76.84 km/h) (as the participants did on many trials—especially toward the end of their experimental session) would not see the white indication of the presence of an object or the change from white to yellow for the advisory. Under these conditions, this warning condition is not useful. The radar requirement for the DAS was that it should only detect objects 350 ft (106.8 km/h) ahead. If the radar requirement increases significantly we will revisit this important problem. In the meantime the use of the yellow advisory is meaningless. Given the 350 ft (106.8 m) radar limitation only the red collision warning was used in the next experiment. This means that the driver saw the white advisory first and then when she or he was 3 s away, the red warning. There is another advantage to use white then red; it is that the contrast between white and red is greater than the contrast between yellow and red. 4.5 SIMULATOR EXPERIMENT 2: USING SPECIALTY VEHICLE OPERATORS TO TEST DRIVER ASSISTIVE TECHNOLOGIES 4.5.1 Introduction and Background The objective of this study was to examine the performance of specialty vehicle operators (Mn/DOT snowplow operators, ambulance drivers, and officers from the Minnesota State Patrol) while driving with the DAS in simulated conditions of limited visibility. Results regarding the lane departure warnings and collision avoidance warnings from Simulator Detailed Design Report 104 August 10, 2001 Experiment 1 were incorporated into this study's experimental design. The study also focused on operator response to lane departure warnings conveyed via the haptic (a vibrating seat) and auditory modalities. For the auditory lane departure warning we used a sound resembling a rumble strip because of the intuitive nature of the sound's message. Rumble strips are commonly employed along road edges to alert drivers of their precarious position on the roadway. Drivers easily recognize the message (i.e., return to the lane) conveyed by the rumble strip and they respond to it quickly. Likewise, an auditory lane departure warning resembling a rumble strip is likely to help drivers maintain their lane position when driving with the HUD. The same reasoning was used in choosing to use the active seat for a lane departure warning. The seat vibrations also resemble the sensations drivers get when crossing rumble strips cut into the road. The experience of driving over a rumble strip varies with: • • • • Rumble strip type Pavement type Vehicle type Vehicle speed We could not hope to reproduce the spectrum of sensations that a driver experiences either aurally through the speakers or haptically through the active seat, so neither the auditory nor haptic lane departure warnings provided an experience veridical to that of physically driving over a rumble strip. Nevertheless, these auditory and haptic lane departure warnings were presented to experimental participants as resembling a rumble strip. The lane departure warnings used in this experiment are highlighted below. As soon as the driver's front tire touched a lane marker, either to the left or to the right of the driver the following warnings were activated on the appropriate side of the display (on the right for a lane departure to the right and on the left for a lane departure to the left): (1) Red-line warning condition on the HUD where the line being crossed changes color from white to red on the HUD. (2) The auditory display emitted a sound resembling a rumble strip (3) The active seat vibrated In this experiment the following warning conditions were used: (1) No HUD (familiarization condition) (2) Lane markings only (3) Lane markings plus visual warning (4) Lane markings plus auditory warning (5) Lane markings plus active seat warning Detailed Design Report 105 August 10, 2001 (6) Lane markings plus visual and auditory warnings (7) Lane markings plus visual and active seat warnings (8) Lane markings plus auditory and active seat warnings (9) Lane markings plus visual, auditory, and active seat warnings (triple modality) All subjects received conditions 1, 2, and 9. Half of the participants received the single modality conditions 3, 4, and 5, and the other half experienced the dual modality conditions 6, 7, and 8. This was done to examine the effects that lane departure warnings presented in isolation have on operator performance and their effects when combined with the warnings from other modalities. 4.5.2 Method Participants Participants were snowplow operators from the Minnesota Department of Transportation (Mn/DOT), ambulance drivers from Hutchinson, and state troopers from the Minneapolis/St. Paul metropolitan area. There were 22 male and 4 female snowplow operators, 11 male and 3 female ambulance drivers, and 16 male state troopers. All were volunteer participants. Apparatus Except as noted below the apparatus for this experiment was identical to that used in Simulator Experiment 1. In this study a sensor was fitted to the turn-signal actuator to determine whether the indicators were activated. The car’s instrument panel and electrical system were powered by a 12 V, 15 A power supply. Acoustic information including lane departure warnings and engine/road noise was generated by the Onyx and fed through a Dolby Digital® receiver (Yamaha #RX-V795a, Tokyo) to four six-inch loudspeakers inside the car (Infinity #63.1i, Tokyo). The front two loudspeakers were mounted in the factory locations in the front doors while the rear speakers were custom-mounted in a traditional rear-deck position. Haptic feedback was transmitted to the driver through the driver’s seat, custom made by JB Research (since acquired by Seat Comfort Systems, Bellflower, CA). The seat had eight discrete vibrators—two on each side of the seat bottom and two on each side of the seat back. (The vibrators on the seat back were not used in this study, however.) These seat bottom vibrators could be fired bi- or contralaterally. The computer governing the systems in the simulator controlled the seat. Detailed Design Report 106 August 10, 2001 A questionnaire was developed specifically for this experiment. Participants were asked to use a visual analog scale to rank the effectiveness of the various warnings or combinations of warnings (see Appendix F). 4.5.3 Experimental Design In the first simulator experiment, lane departure warnings and collision avoidance warnings were investigated. As mentioned in the discussion of the first experiment, the use of a yellow advisory is meaningless until the range of the DAS radar is significantly increased beyond 350 ft (106.8 m). In this experiment only lane departure warnings were investigated. Three different types of display were utilized to deliver the lane departure warnings—(1) the HUD, which visually presented the red line warning (as recommended in the discussion of the first simulator experiment); (2) four speakers mounted in the simulator vehicle, which presented an auditory warning resembling a rumble strip; (3) an active seat, which presented a haptic warning in a way that resembled a rumble strip. These displays were used singly, in pairs, and all three together. It was not possible to present all of these experimental conditions in a single experimental session. In addition, it was not possible to schedule the specialty vehicle operators who participated in this experiment for more than one session. Consequently, the operators were split into two groups that received different sets of lane departure warning conditions. The first group of operators received a set of lane departure warnings that consisted of the three singlemodality warnings along with the triple-modality warning. The second set of operators received a set of lane-departure warnings that consisted of the three dual-modality warnings (along with the triple-modality warning). For the first group of operators the set of single warning conditions (and the triplemodality combination) were randomized using three of the set of six 4 x 4 Latin squares that have the property of completely balancing sequential carry-over effects from one warning condition to another, while none of the rows in the Latin square set is repeated. For the second group of operators the set of dual warning conditions (and the triplemodality combination) were randomized using the other three of that set of six 4x4 Latin squares. The first three trials of the experimental session were the same for every participant— irrespective of which set of warnings to which he or she was assigned. Every session started with a practice trial. This was followed by a trial in which the participant drove in poor visibility conditions without the DAS Interface. Then in the third common trial, the participant drove with the HUD display providing only lane markings. Only after these three common trials did the participants take part in the four trials in which they received the set of single- or dual-modality warnings to which they were assigned. The way in which these design elements were combined is presented in Appendix F in Tables F3, F4 and F5. For each of the specialty vehicle operators who participated in this Detailed Design Report 107 August 10, 2001 experiment, these tables present the order in which he or she experienced the experimental conditions. Specialty vehicle officers were needed as participants for this experiment: Unfortunately they were in short supply. All those who Mn/DOT made available to us were tested. Twenty-six snowplow operators (Appendix F, Table F3) participated in this experiment—two more than were needed to complete a full experimental design cycle of 24 Latin square rows; the two additional snowplow operators were assigned to the first two rows of another cycle. However, only 16 highway patrol officers and 14 ambulance drivers were able to participate. The order for the highway patrol officers is presented in Table F4. The number of highway patrol officers was sufficient to complete four Latin squares—so the within-trials and sequential carry-over balancing was maintained for them. As for the ambulance drivers, the order for the 14 of them who participated is presented in Table F5. Because 14 ambulance drivers participated we could almost complete four Latin squares; the within-trials and sequential carry-over balancing was not quite maintained. This meant that if the results had shown any probabilities that were near the p=0.05 significance level, we would have to interpret them with caution or conduct a more sophisticated analysis (in which the error terms were re-calculated). [You will note in the results section for this experiment that this possibility did not occur.] 4.5.4 Procedure For the experimental trials, the ground was covered with simulated snow—textural features and lane markers denoting boundaries between road, shoulder, and ditch were indistinguishable. The virtual environment was a straight, two-lane highway 3.1 mi (5 km) long with a 10 ft (3.05 m) lane width. Cross streets intersected the road at 1.24 mi (2 km) and 2.48 mi (4 km) and the road had hills 20.3 ft (6.2 m) high at 0.68 mi (1.1 km) and 2.73 mi (4.4 km). The approach and descent for the hills was approximately 983.6 ft (300 m). Vehicles appearing on the road were three-dimensional models with photos of actual vehicles texture-mapped onto the model. The vehicles were positioned at various intervals along the course. The lane position and speed of the vehicles was varied as well with some stationary vehicles placed alongside the road and some moving slowly toward the driver in the opposite lane. After signing the consent form, participants were seated in the simulator to begin a training session to familiarize them with the simulator. The training session had the same format as that employed in Simulator Experiment 1. The first practice trial was a low-visibility driving condition, without snow cover, in which participants were asked to drive through a foggy environment in which the visibility was 98.4 ft (30 m). This trial was used to give drivers an idea of what driving the simulator in poor visibility would be like. The roadway for this trial was a two-lane highway with curves and a few hazard vehicles on the road. After driving for approximately 200 s, the practice trial ended, and the experimenter re-entered the dome to install the combiner for the HUD. Detailed Design Report 108 August 10, 2001 After positioning the combiner in the cabin, the experimenter presented another scenario, used to help the driver calibrate the HUD. This scenario consisted of a two-way stretch of road on which the simulator sat immobilized. Again, visibility was limited. But the low visibility condition was different than that used in the practice trial. Now, the roads were snow covered, so that the lane markings were completely obscured, and there was a fog level of 295.1 ft (90 m)—[this snow-and-fog combination was of similar difficulty to the 98.4 ft (30 m) fog used in the practice (and experiment 1).] The HUD combiner provided a preview distance of 2,459 ft (750 m). Solid white lines were projected onto the combiner to represent the road markings. Participants were instructed to move the combiner until, from their perspective; the lines in the HUD were aligned with the lines on the road. When the combiner was adjusted, the main instructions for the experiment were played to the participant. These instructions, which are presented in Appendix G, provided the participants with background information. Participants were told that, in addition to basic guidance and collision avoidance information provided by the HUD, that lane departure and collision avoidance warnings would be conveyed either through the HUD, through the seat, or through the audio system. The experimenter answered any questions the participant might have had, then, gave the instructions for the first experimental trial. During this trial, the HUD provided lane markings and collision avoidance warnings. Other vehicles were represented in the HUD by a white rectangular outline that appeared when a stationary vehicle was within 350 ft (106.8 m) of the participant. In the experimental trials the participants were instructed to drive at a speed at which they felt comfortable if they were actually driving on a real road with this visibility. They drove the simulated roadway until 2.92 mi (4.7 km) of the course had been covered, when, the experimenter shut down the simulator. Next the participant was given the instructions for the next trial. For the remaining four experimental trials, The HUD provided lane departure warnings, in addition to the lane markings and collision avoidance warnings. The lane departure warnings were conveyed through three different modalities: Visual, auditory, and haptic. The visual warnings appeared in the HUD. When a participant left the lane the line that had been crossed changed from white to red. This particular warning was selected based on results from Simulator Experiment 1. The auditory warning was broadband noise that was digitally manipulated to resemble the sound of a vehicle going over a rumble strip. The auditory warning was presented at an average sound pressure level (SPL) of 70 dBA measured (with a Brüel and Kjær 2260 modular real-time sound analyzer) at head level in the driver's seat. If a participant departed the lane on the right, the warning was given through the right loudspeakers and conversely if the participant departed to the left, the warning was given on the left. The haptic warning was given through the driver’s seat. The warning was meant to simulate the feeling of driving over a rumble strip. When the participant veered out of lane to the left, the vibrators on the left side of the seat vibrated at a frequency of 2 Hz. When the participant deviated to the right, the vibrations fired on the right. All warnings were designed to deploy when the front left wheel of the car crossed over the centerline, or when the front right wheel crossed the fog line. Because the warnings were meant to alert drivers of unintended lane departures, if the participant Detailed Design Report 109 August 10, 2001 activated the turn signal, the warnings were shut down until the signal was deactivated. This was done so that participants could drive around obstructions in their path without experiencing the warnings. Between experimental trials, the experimenter presented prerecorded instructions for the next trial to the participant. Each individual warning or set of warnings had its own set of instructions. A transcript of these instructions is reproduced in Appendix G. When the experimenter was confident that the participant understood the instructions, the next trial began. After finishing all five experimental trials, the participant was asked by a different experimenter to complete the usability questionnaire. Upon completing the questionnaire, the experimenter asked the participant to describe her or his experience and to share any opinions about the system. The participant was then debriefed and thanked for his or her participation. 4.5.5 Results Lane Markings In each experimental session the participants drove five times on a 3.1 m (5 km) course. The visibility conditions were very poor for all five of these drives—the participants drove the course in 295.1 ft (90 m) fog with snow cover obscuring the lane markings. In pre-experimental pilot tests, when pilot subjects drove this course under the same poor visibility conditions—but without the aid of the DAS Interface—it was found that they were unable to stay on the test course. Even driving at very slow speeds—for example, at 5 mph (8.05 km/h)—the pilot subjects could not stay on the road and they drove into the ditch. In contrast, when the experimental participants drove the course under these poor visibility conditions, but using the HUD to provide lane markings, they were able to drive the course. In fact, their average speed increased from trial to trial—this increase is illustrated in Figures 33 through 37. The figures are histograms that show the number of participants that drove at each average speed in each of the five trials. In Trial 1, the HUD only showed the lane markings, whereas in Trials 2 through 5 it presented lane departure warnings (whenever the simulator vehicle moved out of lane) in addition to the lane markings. Detailed Design Report 110 August 10, 2001 SPEED TRIAL: 1 14 12 10 8 Frequency 6 4 Std. Dev = 11.43 2 Mean = 37.4 0 N = 55.00 15.0 25.0 20.0 35.0 30.0 45.0 40.0 55.0 50.0 65.0 60.0 SPEED Figure 33. Histogram showing the mean speeds at which participants drove in Trial 1, in poor visibility conditions, using the lane markings presented on the HUD (with no lane departure warnings). Detailed Design Report 111 August 10, 2001 SPEED TRIAL: 2 16 14 12 10 8 Frequency 6 4 Std. Dev = 10.47 2 Mean = 42.1 0 N = 55.00 15.0 25.0 20.0 35.0 30.0 45.0 40.0 55.0 50.0 65.0 60.0 SPEED Figure 34. Histogram showing the mean speeds at which participants drove in Trial 2, in poor visibility conditions, using the lane markings presented on the HUD (and various lane departure warnings). Detailed Design Report 112 August 10, 2001 SPEED TRIAL: 3 14 12 10 8 Frequency 6 4 Std. Dev = 12.49 2 Mean = 46.5 0 N = 55.00 15.0 25.0 20.0 35.0 30.0 45.0 40.0 55.0 50.0 65.0 60.0 75.0 70.0 SPEED Figure 35. Histogram showing the mean speeds at which participants drove in Trial 3, in poor visibility conditions, using the lane markings presented on the HUD (and various lane departure warnings). Detailed Design Report 113 August 10, 2001 SPEED TRIAL: 4 12 10 8 Frequency 6 4 Std. Dev = 13.51 2 Mean = 47.7 N = 55.00 0 15.0 25.0 20.0 35.0 30.0 45.0 40.0 55.0 50.0 65.0 60.0 75.0 70.0 85.0 80.0 SPEED Figure 36. Histogram showing the mean speeds at which participants drove in Trial 4, in poor visibility conditions, using the lane markings presented on the HUD (and various lane departure warnings). Detailed Design Report 114 August 10, 2001 SPEED TRIAL: 5 20 Frequency 10 Std. Dev = 13.77 Mean = 48.6 N = 55.00 0 20.0 30.0 25.0 40.0 35.0 50.0 45.0 60.0 55.0 70.0 65.0 80.0 75.0 90.0 85.0 95.0 SPEED Figure 37: Histogram showing the mean speeds at which participants drove in Trial 5, in poor visibility conditions, using the lane markings presented on the HUD (and various lane departure warnings). Inspection of the five figures shows that the frequency distributions are approximately normal. Descriptive statistics indicate that the standard deviations of the five distributions are similar. However, as Figure 38 shows, there is an increase in the means of the distributions from the 37.5 mph (60.4 km/h) of Trial 1 to 48.6 mph (8.3 km/h) in Trial 5. The greatest part of the increase occurs between Trials 1 and 3; there is a smaller increase between Trials 3 and 4; and another small increase from Trial 4 to Trial 5. Detailed Design Report 115 August 10, 2001 M ean Speed 55 M ean S peed (m ph) 50 45 40 M ean Speed (m ph) 35 30 1 2 3 4 5 Trial Figure 38. The increase in the mean speed at which the participants drove in poor visibility conditions, using the lane markings presented on the HUD, from Trial 1 to Trial 5. It is clear from Figure 38 that the lane markings provided by the HUD enabled the participants to drive in visibility conditions that otherwise would have made driving extremely problematic. Lane Departure Warnings In this experiment, the lane departure data were handled in a similar manner to the way in which they were handled in the first simulator experiment. As in the first simulator experiment, for all lane departures—except those that occurred within the range within which the DAS radar could determine the presence of an object ahead [i.e., within 350 ft (106.8 m) of a stationary object]—it was determined how long the simulator vehicle remained out of lane. Similarly, the lane departure duration (i.e., the time from the beginning of the lane departure to the moment that the vehicle was completely back in lane) was used as a measure of the effectiveness of the various lane departure warnings. Detailed Design Report 116 August 10, 2001 Also, as in the first experiment, because we found that the distribution of lane departure durations was positively skewed we normalized the data by using a logarithmic transformation. In all, eight ANOVAs were conducted. Four of these ANOVAs compared the durations of the lane departures for those participants who received the single modality warnings and the triple combination—with one ANOVA for each of the three specialty vehicle operator groups, and one that combined the data from all three operator groups. The results of these four ANOVAs are summarized in Table 8. Similarly, the other four ANOVAs compared the lane departure durations for those participants who received the dual modality warnings (and the triple combination)—with one ANOVA for each of the specialty vehicle operator groups, and one for their combined data. The results of these four ANOVAs are summarized in Table 9. [Complete summaries of all eight ANOVAs appear in Appendix D.] Source Subjects (S) Warnings (W) (S x W) Snowplow Operators Highway Patrol Officers Ambulance Drivers 0.013 0.303 0.314 0.000 0.119 0.415 0.098 0.723 0.585 All Specialty Vehicle Operators 0.000 0.528 0.296 Table 8: Single-modality—summary of ANOVAs determining whether lane departure durations were affected by the way in which the warnings were presented (for participants who received the single-modality warnings and the triple modality combination). [More detailed summaries of these ANOVA are presented in Appendix H.] Source Subjects (S) Warnings (W) (SxW) Snowplow Operators Highway Patrol Officers Ambulance Drivers 0.216 0.691 0.146 0.297 0.055 0.891 0.154 0.098 0.648 All Specialty Vehicle Operators 0.016 0.068 0.355 Table 9. Dual-modality—summary of ANOVAs determining whether lane departure durations were affected by the way in which the warnings were presented (for participants who received the dual-modality warnings and the triple modality combination). [More detailed summaries of these ANOVA are presented in Appendix H.] Detailed Design Report 117 August 10, 2001 Tables 8 and 9 show that there were some statistically significant differences between subjects—in particular between the snowplow operators and the highway patrol officers who received the three single-modality and the triple modality lane departure warnings. As mentioned earlier, it is not uncommon to discover between-subject differences. The two tables also indicate that the interactions between subjects and the two warnings types were not statistically significant. Also, and of most interest, there were no statistically significant differences in the lane–departure durations obtained with any of the lane departure warning conditions. Before concluding that the lane departure warnings were of no consequence another comparison should be made—the comparison between the lane departure durations in Trial 1, when no warning was given, and the durations obtained with each of the warning conditions. These comparisons are made in Figures 39 and 40. Average Lane Departure Durations Single M odality 4 Am bulance Single Highw ay Patrol Single S nowplow Single 3.5 S econds Out of Lane 3 2.5 2 1.5 1 0.5 0 1 2 3 4 8 W arning # Figure 39. Average (antilog of mean log of duration) lane departure durations for snowplow operators, highway patrol officers, and ambulance drivers who received the single-modality warnings and the triple modality combination. [Key to Warning Conditions—W # 1 is No Warning; W # 2 is Visual Warning; W # 3 is Auditory Warning: W # 4 is Haptic Warning; W # 8 is Triple-Modality Warning] Detailed Design Report 118 August 10, 2001 Average Lane Departure Durations D ual M odality 4 Am bulance Dual Highw ay P atrol Dual S nowplow Dual 3.5 S econds Out of Lane 3 2.5 2 1.5 1 0.5 0 1 5 6 7 8 W arning # Figure 40. Average (antilog of mean log of duration) lane departure durations for snowplow operators, highway patrol officers, and ambulance drivers who received the dual-modality warnings and the triple modality combination. [Key to Warning Conditions—W # 1 is No Warning; W # 5 is Visual plus Auditory Warning; W # 6 is Auditory plus Haptic Warning: W # 7 is Haptic plus Visual Warning; W # 8 is TripleModality Warning] Detailed Design Report 119 August 10, 2001 Figure 39 and Figure 40 show, for both groups of highway patrol officers and both groups of ambulance drivers, and for the group of snowplow operators who received the dual-modality warnings, that the average lane departure durations were at least 0.5 s less when they were given lane departure warnings than they were not given a warning. The only exception to this general finding was for the group of snowplow operators who received the single-modality warnings—and for this group, the average lane departure duration, when they were not given a warning, was considerably smaller than for the other five groups. Collision Avoidance Warnings In the first simulator experiment it was noted that when subjects were driving in poor visibility conditions and were approaching stationary vehicles at the side of the roadway they crossed the centerline 68% of the time when a truck-size white rectangular outline was used to indicate its presence and location and no other warning was given. When collision avoidance warnings were also given this percentage increased to an average of 83% (the percentages were 82% for one warning group and 84% for the other). (The subjects used in the first simulator experiment were members of the general public, and not specialty vehicle operators). In this experiment with specialty vehicle operators there was only one warning condition. In all five trials, when the participant approached a stationary vehicle ahead—either half in the lane (and half on the hard shoulder) or just outside the edge-line—that vehicle was outlined in white when the participant was 350 ft (106.8 m) from it. Then, when he or she was 3 s away, the outline changed from white to red. The specialty vehicle operators responded by crossing the centerline far more often than the participants in the first experiment. Table 10 gives a breakdown of the responses for the snowplow operators, highway patrol officers, and ambulance drivers, as well as the combined result for all the specialty vehicle operators and, for comparison purposes, the percentage for the participants of the first experiment. Type of Operator Snowplow Operators Highway Patrol Officers Ambulance Drivers All Specialty Vehicle Operators Operators in Experiment #1 (General Public) Percent of Lane Crossing Responses 92.9 % 92.3 % 99.0% 94.3% 83 % Table 10. Percentage of Lane–crossing responses for the three types of specialty operator, individually and combined, and (for comparison purposes) the subjects in Experiment #1. Detailed Design Report 120 August 10, 2001 Questionnaire Data Copies of the questionnaires given to the participants in the single mode and double mode groups are presented in the Appendix. After each question, a response bar was presented. At each end of this response bar anchor points were presented. These anchor points reflected the extremes of the possible responses for each question. The participants were asked to mark the bar in a location that indicated their response. Each response was measured from the left to the mark made by the driver; then these measurements were rescaled from zero to 100. Resultant scores near zero would be very unfavorable, those near 50 would be neutral, and those near 100 would be very favorable. The first five questions dealt with the usefulness of the DAS Interface and the information that was presented by it. These five questions were identical for all participants. The means presented Table 11 are averaged across all participants. Question Overall Mean 1. How useful were the lane markings provided by the HUD? 2. How useful was it for the HUD display to let you know there were vehicles ahead? 3. How useful were the collision avoidance warnings? 4. How easy was it to drive in the poor visibility conditions using the HUD? 5. How useful were the lane departure warnings? 87.9 91.7 81.2 82.7 78.2 Table 11. Average response to questions about the DAS Interface and the information presented by it. [Responses could vary between zero and 100—a score close to zero indicates an unfavorable response; a score close to 100 indicates a favorable response.] As Table 11 indicates, the participants found that the lane markings provided by the HUD were very useful. Similarly, they found it very useful that the HUD informed them about the presence of vehicles ahead. They found that the collision avoidance warnings were useful, and that it was relatively easy to drive in poor visibility conditions when using the HUD. They also found that the lane departure warnings were useful. The four remaining questions dealt with the specific lane–departure warning conditions that the participants experienced. Because they did not all receive the same set of warning conditions the responses to these four questions are given in two tables. The responses of the participants who received the single modality lane departure warnings (and the triple combination) are given in Table 12, while those who received the lane departure warnings in two modalities (and the triple combination) are given in Table 13. Detailed Design Report 121 August 10, 2001 Question Overall Mean 6. How useful as a lane departure warning was the red line? 7. How useful as a lane departure warning was the rumblestrip sound? 8. How useful as a lane departure warning was the vibration in the driver’s seat? 9. How useful as a lane departure warning was the combination of the red line, rumble-strip sound and vibration in the driver’s seat? 69.2 61.7 69.9 77.8 Table 12. Average response to questions about the lane departure warnings from the participants who received the warnings via three single modalities and by the combination of all three modalities. [Responses could vary between zero and 100—a score close to zero indicates an unfavorable response; a score close to 100 indicates a favorable response.] Question Overall Mean 6. How useful as a lane departure warning was the combination of the red line and the rumble-strip sound? 7. How useful as a lane departure warning was the combination of the rumble-strip sound the vibration in the driver’s seat? 8. How useful as a lane departure warning was the combination of the vibration in the driver’s seat and the red line? 9. How useful as a lane departure warning was the combination of the red line, rumble-strip sound and vibration in the driver’s seat? 75.2 72.2 81.8 74.4 Table 13: Average response to questions about the lane departure warnings from the participants who received the warnings via dual modalities and by the combination of all three modalities. [Responses could vary between zero and 100—a score close to zero indicates an unfavorable response; a score close to 100 indicates a favorable response.] Table 12 shows that the participants who received the single modality lane departure warnings gave lower usefulness ratings to each of the three single modality warnings than they did to the triple combination. In contrast, Table 13 shows that the participants gave usefulness ratings to two of the dual modalities lane departure warnings (the visual plus Detailed Design Report 122 August 10, 2001 auditory, and the auditory plus haptic) that were similar to the ratings they gave to the triple combination (although it should be noted, they gave even higher ratings to the remaining dual-modality combination (the haptic plus visual). 4.5.5 Discussion Lane Markings Results from this experiment in which the participants were professional Specialty Vehicle Operators concur with findings from the first experiment that used drivers selected from the general public. Both experiments show that by using lane marking information provided by the HUD, it is possible to drive in visibility conditions in which it would otherwise be extremely difficult if not impossible to drive. Figure 38 shows that the first time the Specialty Vehicle Operators drove with lane markings provided by the HUD their average speed was 37.5 mph (60.4 km/h). The figure also shows that after three brief 3.1 mi (5 km) drives in Trials 1 through 3 the average speed of the operators increased by 10 mph (16.1 km) to 47.7 mph (76.8 km/h). These speeds are probably faster than would be safe if the visibility conditions in real life were as bad as those simulated here. In this experiment the operators encountered a combination of 295.1 ft (90 m) fog and a white ground cover—simulating snow—that obscured all lane markings. Therefore, with regard to visibility, the simulated conditions were as bad, or perhaps even worse, than anything that the operators encounter in real life operations. However, it should be noted that while it looked to the operators as if they were driving on snow, it did not feel as if they were. Effectively the contact between the simulator vehicle and the road was the same in this experiment as if they had been driving on dry pavement—and, in addition, there was no build up of snow on the windshield as there would be in real life. Lane Departure Warnings Table 8 indicates that there were no statistically significant differences in lane departure duration for the three single-modality warnings and the triple-modality combination; similarly Table 9 indicates a similar result for the three dual-modality warnings and the triple-modality combination. However, Figure 34 and Figure 35 suggest that the conditions in which a lane departure warning was given (irrespective of which one it was) resulted in shorter lane departure times than when no warning was given. It would appear that giving a warning is better than not giving a warning. The results of the questionnaire suggested that the participants who received the singlemodality warnings thought that the triple-modality warning was more useful (see Table 12). The questionnaire responses for those who received the dual-modality warnings were similar for the triple-modality warning—although these participants did give a higher rating to the combination of the haptic and visual warnings (see Table 13). Detailed Design Report 123 August 10, 2001 Collision Avoidance Warnings As Table 10 indicates, the presence of vehicles partially in, or to the side of the road produced considerably higher percentages of lane-crossing responses from all three types of specialty vehicle operators than their presence did from representatives of the general public (the subjects in the first simulator experiment). In their questionnaire responses (Table 11), the Specialty Vehicle Operators also indicated that they found it very useful that the HUD informed them about the presence and location of vehicles ahead, and that they found that the collision avoidance warnings were useful. 4.6 ROSEMOUNT FIELD STUDY: DRIVING A SNOWPLOW ON A CHALLENGING TEST TRACK WHILE USING THE DAS IN CONDITIONS OF ZERO VISIBILITY 4.6.1 Introduction and Background The objectives of this study were twofold: (1) to determine whether expert snowplow operators could drive in conditions of zero visibility using the DAS interface developed by implementing findings from the two driving simulator experiments and (2) to learn whether they were comfortable with the DAS interface. This meant we planned to have them drive on narrow roads with the view of the outside environment completely occluded. Their only source of information was the lane markings on the HUD and the lane departure and collision avoidance warnings. There are at least two ways to conduct a field study of this nature. The first is modeled on traditional laboratory experiments in which a full factorial design is employed. This approach requires extremely robust equipment. The second approach involves acquiring knowledge from experts, sometimes with a change to the conditions in which they are performing. Then the experts are asked to give concurrent (rather than retrospective) verbal descriptions of what they are doing. Unfortunately, the field equipment was not sufficiently robust for the kind of rigor required for a full factorial experimental design that could be done within the time period available. But fortunately, because our subjects were expert snowplow operators, we were able to use the second approach. The development of the knowledge acquisition technique we used is described below. Verbal reports had been largely discounted as a credible source of research data, until the early 1980s which saw the beginning of an extensive use of verbal data to study cognitive processes, with many researchers using protocol analysis developed by Ericsson and Simon (1980). Concurrent verbal reports are now generally recognized as major sources Detailed Design Report 124 August 10, 2001 of data on cognitive processes in specific tasks (Anderson, 1987), and verbal reports are a primary data source in many domains including decision making (Montgomery & Svenson, 1989), knowledge acquisition for photo interpreters (Shalin, Bloomfield, Bullemer, Bhanu, & Shelton, 1989), pilot decision making (Bloomfield, Peio, Lehman, Masters, & Boettcher; 1989), user testing of computer products (Denning, Hoiem, Simpson, & Sullivan, 1990), expert knowledge acquisition of pilots in take-off and landing (Bloomfield, Shalin, & Corwin, 1991), and system development and instructional design (Hoffman, Crandall, & Shadbolt, 1998). The particular method used in this study has its basis in the work by Hoffman (1984). He elicited tacit knowledge from experts (photo interpreters) while they worked. He altered the working conditions of the task (in this case artificially reducing the time available). Shalin, et al (1989) followed this with a different alteration to the working conditions. Specifically, they changed the visual frame of reference for photo interpreters. The notion of changing the task environment was developed further by Bloomfield and Shalin (1989) who suggested systematically manipulating task conditions on a trial-bytrial basis. They (Bloomfield, Shalin, & Corwin, 1991) later used this technique to acquire knowledge from MD80 pilot instructors engaged in takeoff and landing in a motion-based flight simulator. In this study they systematically eliminated key information sources and procured extensive tacit information/knowledge that the instructors were unaware they needed to pass on to their students. As mentioned above, in this study we planned to have the snowplow operators drive with the outside view completely occluded with the only source of information provided by the HUD, lane departure, and collision avoidance warnings. This essentially inverts the Bloomfield and Shalin knowledge sensitive task manipulation process because the outside view is occluded rather than the inside view of the environment. In this study we acquired information about the design of the DAS from the snowplow operators while they drove the test course without and with occluding the external environment. 4.6.2 Method Participants Participants were 13 Mn/DOT snowplow operators from the Minneapolis/St. Paul metropolitan area. Their participation was voluntary. Detailed Design Report 125 August 10, 2001 Apparatus Visual navigational information inside the snowplow was conveyed via a Head-Up Display (HUD) developed by the Intelligent Vehicles Lab, at the University of Minnesota. Images on the HUD were projected from a 10.4 in (266 mm) Mobile VU projector (Litton Systems #46830-1, San Diego) mounted just to the right of the driver's headrest. The roadway markings were generated by the Differential Global Positioning System. Each subject saw the lane markings (provided by DGPS via the projector) on the HUD. The data acquisition system used for the field tests was a variation of the vehDAQ system to be used for the field operational test proper. The block diagram of the system is provided in Figure 41 below. Functionally, it operates identically to the vehDAQ system described in the following chapter with the exception that video data is written to a digital tape rather than compressed and written to a computer hard drive. This was done for two reasons. First, the experiments run at Rosemount lasted less than three hours, and that data would fit on a standard digital video tape. Second, the vehDAQ had not been fully developed, and therefore it was unavailable for use. Engineering unit data was synchronized with the video frames, and stored on the vehicle computer shown below. Audio 1 Camera 1 Camera 2 Video Quad Unit Titler Digital VCR Camera 3 Camera 4 time stamp for synchronization engineering unit data from vehicle sensors Vehicle computer Figure 41. Block diagram of the data acquisition system used in the Rosemount field Study Detailed Design Report 126 August 10, 2001 The combiner surface was vertically and horizontally planar convex. In normal use when the driver looked forward he or she was able to see both the imagery on the combiner and the outside environment. (In this experiment, on all but the familiarization trial, the view out of the windshield and side windows was occluded.) The combiner was mounted on an adjustable frame so that it could accommodate drivers of different heights. The combiner measured 11.5 in (294.9 mm) wide and 6 in (153.8 mm) high and was framed with clear 0.75-inch (19.2 mm) polycarbonate. Test Site The experiment was conducted on a test track at the University of Minnesota's Agricultural Research Extension Station in Rosemount. The track, shown in Figure 42 below, was 4 miles long. The track consisted of three legs: the first was a two-lane-wide road; the second was a three-lane-wide road; and the last leg was a single lane. All three were unpaved. In the first leg soon after the start point there was a gentle left bend. Between the first and second legs there was an oblique approximately 120 deg left turn. Between the second and third legs there was a 90 deg short radius left turn. Then in the middle of the third leg there were two 90 deg short radius turns, the first to the left and the second to the right. The experiment was run in December 2000 and January 2001 a time during which the temperature varied between 30 deg and minus 18 deg F with wind chills that were sometimes as low as minus 50 deg F. On all test runs the track was snow covered. If necessary the track was plowed at the start of each day. Atmospheric visibility was excellent on all days. 4.6.3 Design and Procedure On arrival at the Rosemount facility each participant signed an informed consent form. They drove from the assembly point to the start of the course. Subjects were asked to adjust the combiner by centering its lane markings in the middle of the road. He or she also made an azimuth adjustment to assure that the vanishing point of the HUD lane markings coincided with the vanishing point of the road. Then they listened to taped instructions. The transcript of the instructions is located in Appendix I. Before starting the first test run the experimenters checked with the traffic controllers to ensure that no other traffic was present on the course and to block any traffic from turning onto it. Following clearance from all six traffic control points, shown in Figure 42, the subject began driving. Detailed Design Report 127 August 10, 2001 Figure 42. The Rosemount test track showing the course and the six checkpoints for traffic control Run Number 1: During the first run the drivers were asked to pay attention to the lane markings on the HUD through which the actual roadway was clearly visible. They also were invited to sample the lane departure warnings. At Turn Number 1, which is an approximately 120 deg oblique angle they turned left onto the next stretch of road. During the turn subjects were instructed to attend to the HUD and to their speed. They were told that during the conditions in which their view of the actual road would be completely occluded the lane markings on the HUD would disappear and that they "should proceed to turn cautiously at a speed that did not exceed 2 to 3 mph." (The reasons for this are discussed below.) At turn 2, which was a sharp 90 deg turn to the left, onto a single lane road each subject was again cautioned to turn slowly and cautiously because on sharp turns the HUD's lane markings completely disappeared. They were told that the best way to navigate the curve was with a foot over the brake and not on the accelerator. This guidance was useful to them in preparation for test runs during which their view of the actual roadway was occluded by opaque curtains covering the inside of the windshield and side windows. They were told that when negotiating the sharp 90 deg turns they would need to rely on Detailed Design Report 128 August 10, 2001 the auditory and active seat warnings for guidance. At turns 3 and 4, a sharp 90 deg left turn and sharp 90 deg right turn, respectively, the subject was again cautioned to proceed around the turns "at a speed of approximately 3 to 4 mph." They were again told that the lane markings on the HUD would disappear on the tight turns. At the end of the course they were asked to stop at the point on the track where they would be asked to stop on subsequent runs. Subjects were encouraged throughout the run to comment on their perception of the HUD and lane departure warnings. Runs Number 2 and Number 3: During the second and third runs subjects were invited to drive with opaque curtains drawn across the windshield and side windows. They were asked to drive the course by looking at the road markings on the HUD. As in run Number 1 subjects were encouraged throughout the run to comment on their perception of the HUD and lane departure warnings. Runs Number 4 and Number 5: In the fourth and fifth runs the collision avoidance warnings were activated. Virtual objects were used due to the potential for crashes if real objects had been placed on or alongside the road. Subjects were given information about the how the collision avoidance warnings worked. As in runs 1, 2, and 3 subjects were encouraged to comment on their perception of the HUD and lane departure warnings. In runs 4 and 5, however, they were also asked to comment on the collision avoidance warnings. 4.6.4 Results and Discussion The field test was treated as an expert knowledge extraction task. As mentioned earlier, the test sessions began with a trial run in which each snowplow operator familiarized himself or herself with the test track. Then, for the remaining runs, the operator drove with the outside view of the environment completely occluded, so that their main source of external information was the DAS Interface (i.e., the HUD, the auditory display and the active seat). The operators were driving under conditions that essentially inverted the knowledge sensitive task manipulation process used by Bloomfield et al. (1991) with the outside view of the environment occluded rather than portions of the internal information/control panel environment. This study also differed from previous expert knowledge extraction studies in that it focused on the operators’ opinions about new technology that was used under an extreme condition. Their opinions are discussed below. The discussion includes the results of statistical analyses that were conducted when the available data met the test requirements. The discussion begins with the operators’ overall impressions of the Driver Assistive System itself, and then continues with their comments on specific aspects of the DAS Interface. Detailed Design Report 129 August 10, 2001 General Impressions Generally, the snowplow operators thought that the DAS was a good system (p<0.006, Binomial test; Siegel & Castellan, 1988). Many thought that it would be useful if deployed on the snowplow that they usually drive, especially those who have had significant experience with plowing snow in Western Minnesota where severe whiteout conditions due to blowing snow are not uncommon. Operators who have plowed exclusively in the Twin Cities metropolitan area also were impressed with the DAS, but indicated that they rarely, if ever, experience the whiteout conditions under which it would be particularly useful. In general the participating operators thought that others would be able to use the DAS. After some initial trepidation about driving by using the lane markings presented on the HUD, the operators were able to drive using them when the outside view was completely occluded (p<0.00001, Binomial test; Siegel & Castellan, 1988). And, in general they became more comfortable and confident in later trial runs. They did mention that they would like a greater distance between the combiner and the projector. Lane Departure Warnings With regard to the visual lane departure warnings these were generally found to be useful. They found that the message delivered by the red line warning to be self-evident and useful. They said that they would like to see it in the final version. With regard to the auditory warning and the haptic warning that was delivered via the active seat, it was clear that the two warnings were perceived by the operators as a single perceptual unit; in some cases, the operators were clearly unaware that the signals were separate at all. It is likely that the signal presented by the two modalities merged into a single perceptual entity because they both oscillate. The oscillating quality also sets them apart from the constantly presented red line visual warning. During their fourth and final trial run, three operators chose to drive the course with only one from this pair of signals. After doing this, all three reported that using only one of the pair considerably diminished the effect of the combined warning, while the combination accentuated it; they recommended that the DAS should be used with the combination. It is clear that both signals should be used concurrently. In the second simulator experiment many snowplow operators suggested that the auditory warning they experienced in the laboratory experiment would be lost in the noisy cab of the snowplow. In contrast, in the field study where the auditory signal was always experienced as part of a combined signal, none of the participants ever suggested that the combination of the auditory and active seat warnings would be a problem when deployed. Detailed Design Report 130 August 10, 2001 Turn Advisories Unfortunately when sharp turns are negotiated, the lane markings on the HUD and the red line warning both disappear, and not surprisingly the operators indicated that they had problems with the way in which turns were depicted on the HUD (p=0.061, Binomial test; Siegel & Castellan, 1988). This is because the lane markings shown on the HUD match the normal view that a driver has out of the windshield view, but that this conformal view fails when the vehicle is negotiating sharp turns. When negotiating turns, all drivers need to look out the side window at some point in order to successfully complete the turn. At the moment, we have not determined how shallow a turn must be before it is possible to use only information obtained through the windshield. However, it is clear that for the 120 deg turn as well as the three 90 deg short radius turns that the snowplow operators experienced when driving on the test track, they need to look out the side window to complete them. The everyday driver may believe that he or she acquires the information necessary for taking sharp corners through the windshield and never through the side windows—he or she is clearly wrong. The reason for such an erroneous belief may be that driving is a highly over learned and over practiced skill, that expert knowledge is difficult if not impossible for the expert to articulate, and that there is, as Howarth (1988) points out, “a surprising degree of dislocation […] between conscious verbally expressed behavior and tacit knowledge which guides skilled behavior.” With regard to the DAS, a conformal view of the lane markings is clearly to be recommended when the operator is driving on straight road segments. If there is no change to this display (to, for example, a plan-view) when turns are to be negotiated, the information needed from the side windows must be provided in some other way. We found in the field test that the operators were able to negotiate the turns using the combined auditory/active seat signal as a turn advisory (p<0.00001, Binomial test; Siegel & Castellan, 1988). Some operators experienced difficulties with completing the turns, particularly on the earlier test runs—and this finding has implications for operator training. Collision Avoidance Warnings The operators thought that the white truck-sized outline indicating the presence of an object ahead was good. They also liked the use of the change in color of this outline from white to red as a collision avoidance warning. However, they expressed considerable doubt that the time at which this warning was given (i.e., 3 s before a collision would occur) would be sufficient to take corrective action. This is in part because the dynamics of a moving snowplow might require a longer period for the operator to avoid a collision. As a result, we believe that further research to determine an appropriate collision warning interval is necessary. Detailed Design Report 131 August 10, 2001 4.7 DRIVER ASSISTIVE SYSTEMS DESIGN RECOMMENDATIONS This section concludes with some DAS Interface design recommendations that are based on considerations of the findings from investigations carried out in the two driving simulator experiments and the field study, and how they modify previous DAS Interface recommendations (from the Technical Systems Requirements, July 2000). 4.7.1 Lane Markings As expected, the two simulator experiments demonstrated that when lane markings derived from DGPS are presented on the HUD they aid specialty vehicle operators in the task of driving in conditions of extremely poor visibility. Further, the field test showed that snowplow operators could drive the straight stretches on a test track and narrow lanes in conditions of zero visibility. As a result, it is clear that lane markings presented on the HUD should be a recommended part of the DAS Interface. 4.7.2 Lane Departure Warnings & Turn Advisories Four ways of providing visual lane departure warnings were compared in the first simulator experiment. When compared to a no-warning control condition, all of these visual warnings led to a statistically significant reduction in the duration of lane departures. There was, however, no significant difference in the lane departure durations that were associated with the four different ways in which the visual warnings were delivered. Because of this other criteria were used to decide between them. Two were rejected because they could partially obscure the driver’s view of the road surface as well as objects (including vehicles) that might be present in the area that they were moving into. A third was rejected because it could be confused with an existing standard warning. The fourth visual lane departure warning entails a change of color of the line on the HUD that marks the lane boundary that the operator’s vehicle is inadvertently crossing. The color changes to red. Because the red line warning was effective, visually salient, and consistent with the way in which red is used in the MUTCD (2000), it is recommended for use as the visual lane departure warning on the HUD. In the second simulator experiment, auditory and haptic lane departure warnings were investigated along with the red line visual warning from Experiment 1. The three warnings were used alone, in combinations of two, and all three together. As in the first simulator experiment, when compared to a no-warning condition, the conditions with a lane departure warning were associated with lower average lane departure durations. Questionnaire data obtained from the specialty vehicle operators who were the participants in this experiment suggested that the combination of warnings presented simultaneously in all three modalities might be the most useful—although there might be problems in actually hearing the auditory signal in the cab of a snowplow. However, when the triple modality lane departure warning was used in the field test, it appeared that the oscillating nature of the auditory and haptic caused them to merge Detailed Design Report 132 August 10, 2001 together so that they were perceived as a single perceptual entity by some of the snowplow operators. Also, in the field test when the snowplow operators were negotiating 90 deg short radius turns, it transpired that inevitably the visual lane markings and the red line lane departure warnings could no longer be seen on the HUD. When this happened in the zero visibility conditions in which the field test was conducted, the only way that the 90 deg short radius turns could be negotiated was by driving very slowly—i.e., in the 2 mph (3.22 km/h) to 4 mph (6.44 km/h) range—and by using the combined auditory/haptic signal as a turn advisory. It is to be expected that this combination will be very useful for negotiating sharp turns when snow cover is obliterating ground features and/or in extremely poor visibility conditions. The recommendation with regard to lane departure warnings and turn advisories is that the triple modality combination should be used. 4.7.3 Indicating the Presence of an Object Ahead To indicate the presence of objects ahead that could become obstacles in the path of the specialty vehicle operator, we initially recommended the use of a white rectangular outline that is the width of the widest vehicle that a driver is likely to encounter—i.e., a truck. In addition, we recommended that the width of the rectangle should be proportional to the width of the cross-section of a truck at the distance that the object is from the driver, so that the size of the rectangular outline will increase proportionately (and dynamically) as the operator approaches it. It was also recommended that the rectangular warning symbol should only be outlined—so that when the operator gets close enough to see the details of the object within the open rectangle, it will be possible to determine what it actually is—whether it is indeed a truck or whether it is a smaller vehicle. When answering the questionnaire presented to the specialty vehicle operators that participated in the second simulator experiment, the question that received the highest usefulness rating (91.7—see Table 11, question 2) asked “How useful was it for the HUD display to let you know there were vehicles ahead.” And in the field test, the snowplow operators indicated that it was useful to indicate the presence of a vehicle ahead by using the white truck-sized rectangular outline. Our recommendation remains that the presence of an object ahead should be indicated by a rectangular white outline proportional to the width of the cross-section of a truck at the distance that the object is from the driver. Detailed Design Report 133 August 10, 2001 4.7.4 Collision Avoidance Warnings In the first simulator experiment, two collision warning conditions were investigated. These warnings were initiated when the object ahead, in fact, was an obstacle that partially blocked the operator’s route. And, in both warning conditions the color of the rectangular outline changed. In one warning condition, an advisory was given when the distance between the operator and the obstacle became 5 s or less (then the rectangular outline changed in color from white to yellow), and a warning was issued when the distance between the operator and the obstacle became 3 s or less (then the rectangular outline changed from yellow to red). In the second condition, no advisory was issued; instead, there was only a warning when the distance between the operator and the obstacle became 3 s or less (in this case, the rectangular outline changed from white to red). In the first experiment, the advisory was ineffective. But this was because in many trials the subjects were traveling above 47.72 mph (76.84 km/h), and anyone traveling at speeds higher than that will have covered 350 ft (106.8 m)—the maximum range at which the DAS radar acquires objects—in less than 5 s. In trials where this occurred, no change in the color of the rectangular outline from white to yellow could have occurred, because when it first appeared the outline would have already been yellow. Because of this the yellow advisory/red warning combination was not used in the second simulator experiment or the field test. However, if a radar with a longer range than 350 ft (106.8 m)—is acquired for use in the DAS, this advisory condition should be re-considered. The second warning condition in which a warning was issued when the distance between the operator and the obstacle became 3 s or less (where the rectangular outline changed from white to red) appeared to be more useful. Its use increased the percentage of trials in which the operators moved over the centerline when approaching a stationary object that was partially blocking, or only just out of, the lane ahead from 68% (with no warning) to 82% (with the red warning). Then, when it was reused in the second simulator experiment, the specialty vehicle operators moved over the centerline when approaching a stationary object that was partially blocking, or only just out of, the lane ahead over 90% of the time when the red collision warning was issued. Finally, when the use of the red collision avoidance warning was demonstrated to the snowplow operators who participated in the field test, they indicated that they thought it would be useful. As a result of this, we recommend that the change in color from white to red of the rectangular outline indicating the presence of an object be used as a collision avoidance warning when an object becomes an obstacle in the path of a specialty vehicle operator. However also in the field test, the snowplow operators expressed considerable doubts about whether a 3s warning would be sufficient for them to take corrective action to Detailed Design Report 134 August 10, 2001 avoid a collision. As a result, we believe it is necessary to conduct further research to determine a perhaps more appropriate collision warning interval. 4.7.5 Magnetic Tape Vehicle Positioning System As part of the vehicle sensing suite, the magnetic lateral lane position is used as a redundant sensor to the GPS based positioning system. As was described in the previous chapter, the magnetic sensor provides a measure of the distance to a tape embedded in the roadway; should GPS become unavailable, it provides a reference for estimating vehicle position without GPS for a short period of time. When the estimated vehicle position error grows too large, the vehicle position estimate is no longer useful, and the only reliable information becomes the distance to tape measurement. The essential question is that of preview versus no preview. The DGPS/preview combination provides a forward view down the road that, even in very poor visibility conditions, would allow a driver to travel at considerable speed. In contrast, the information from the magnetic tape vehicle positioning system only indicates the current position of the vehicle relative to that lane marker. It is essentially a guidance system that is useful when the operator can use environmental cues to maintain speed but cannot see lane markings because they are snow covered. It provides only current position, but no preview. Preview alone does not enable the driver to travel at speeds comparable to the DGPS/preview condition. In very poor visibility conditions if an operator were traveling at the speeds allowed by the DGPS/preview combination, if that combination were to abruptly fail, the operator would have to reduce speed, and then switch the vehicle guidance task using only lateral displacement information. That sequence of events would not allow for a smooth transition. Considerable thought must go into what would make for a smoother transition. 4.7.6 Other Observations from the Specialty Vehicle Operators In addition to the design recommendations derived from the two driving simulator experiments and the Rosemount field study, other observations gleaned from the specialty vehicle operators should be considered. They are presented below: • A number of snowplow operators expressed the need for bridge advisories on the HUD. Each year across Minnesota a number of boxes are knocked off the backs of snowplows. This is an expensive loss and one that would likely be reduced if drivers had bridge overpass information on the HUD. • Ambulance drivers and highway patrol officers indicated that they would like to see street markings on the HUD. The street markings would help them to more easily reach their destination when visibility is limited. Detailed Design Report 135 August 10, 2001 • A number of snowplow operators observed that the DAS would be very useful in outstate Minnesota but probably would not be needed in the Metropolitan area. • Some operators, but not all, suggested it might be useful for gang plowing. • Finally, most snowplow operators expressed concern about the close proximity of the combiner and the projector to their head. This safety concern should be resolved if the system is to be widely deployed. 4.7.7 References Anderson, J.R. (1987) Methodologies for studying human knowledge. Behavioral and Brain Sciences, 10, 467-505. Bloomfield, J.R., Peio, K.J., Lehman, E.F., Masters, R.M., and Boettcher, K.L. (1989) Flight decision characterization. Draft Final Report (63578-89U/P6754) for Air Force Systems Command, Wright-Patterson AFB, Dayton, Ohio. Bloomfield, J.R. and Shalin, V.L. (1989) Knowledge acquisition techniques: Problems and Potentialities. In: Megaw, E.D. (Editor) Contemporary Ergonomics 1989. London: Taylor & Francis, 164-171. Bloomfield, J. R., Shalin, V. L., and Corwin, W. H. (1991). Knowledge-sensitive task manipulation: acquiring knowledge from pilots flying a motion-based flight simulator. In: Jensen, R. S. (Editor) Proceedings of the 6th Aviation Psychology Symposium, Columbus, OH, 1068-1073. Box, G.E.P. (1954a) Some theorems on quadrative forms applied in the study of analysis of variance problems; I: Effect of inequality of variance in the one-way classification. Annals of Mathematical Statistics, 25, 290-302. Box, G.E.P. (1954a) Some theorems on quadrative forms applied in the study of analysis of variance problems; II: Effects of inequality of variance and of correlation between errors in a two-way classification. Annals of Mathematical Statistics, 25, 484-498. Denning, S., Hoiem, D., Simpson, M., & Sullivan, K. (1990) The value of thinkingaloud protocols in industry: A case study of Microsoft. Proceedings of the Human Factors Society—34th Annual Meeting, Volume 2. Santa Monica, California: The Human Factors Society, 1285-1289. Ericsson, K. A. and Simon, H.A. (1980) Verbal reports as data. Psychological Review, 87, 215-251. Detailed Design Report 136 August 10, 2001 Färber, B. (1991) Designing a distance warning system from the user’s point of view. APSIS Report. Glonn-Haslach, Germany: Institut für Arbeitspsychologie und Interdisziplinaire Systemforschung. Ferguson, G.A. (1959) Statistical Analysis in Psychology and Education. New York: McGraw-Hill, Inc. Fisher, R.A. (1935) The Design of Experiments. Edinburgh and London: Oliver & Boyd. Hawkins, R.K. (1988) Motorway Traffic Behaviour in Reduced Visibility Conditions. In: Gale, A.G., Freeman, M.H., Haslegrave, C.M., Smith, P., & Taylor, S. (Editors) Vision in Vehicles—II. North-Holland: Elsevier Science Publishers, B.V., 9-18. Hoffman, R.R. (1984) Methodological preliminaries to the development of an expert system for aerial photo interpretation. US Army Engineer Topographic Laboratories: Technical Report No. ETL-0342. Hoffman, R.R., Crandall, B. & Shadbolt, N. (1998) Use of the Critical Decision Method to Elicit Expert Knowledge: A Case Study in the Methodology of Cognitive Task Analysis. Human Factors, 40(2), 254-276. Howarth, C.I (1988) The relationship between objective risk, subjective risk, and behavior, Ergonomics, 31, 527-535. Maretzke, J. & Jacob, U. (1992) Distance warning and control as a means of increasing road safety and ease of operations. In: FISITA ’92: Safety, the Vehicle, and the Road, XXIV FISITA Congress. London: Institute of Mechanical Engineers, 105-114. Michon, J.A., Piersma, E.H., Smiley, A., Verwey, W.B., & Webster, E (1990) Design considerations. In: Michon, J.A. (Editor) Generic Intelligent Driver Support. London, U.K.: Taylor & Francis, 69-87. Montgomery, H. & Svenson, O. (1989) Process and Structure in Human Decision Making. Chichester, England: John Wiley & Sons. MUTCD (2000) Manual on Uniform Control Devices (Millennium Edition). Department of Transportation. Washington DC: Federal Highway Administration. U.S. Neider, J., Davis, T., & Woo, M. (1993) OpenGL Programming Guide. Reading, Massachusetts: Addison-Wesley Publishing Company. Ng, L. and Barfield, W, (1998) Determining user requirements for intelligent transportation system design. In: Barfield , W. and Dingus, T.A. (Editors) Human Factors in Intelligent Transportation Systems. Mahawah, New Jersey; Lawrence Erlbaum Associates, Inc., 325-357. Detailed Design Report 137 August 10, 2001 Norman, D. (1998). The Design of Everyday Things. London: MIT Press. Shalin, V.L., Bloomfield, J.R., Bullemer, P., Bhanu, B., and Shelton, C. (1988). Cognitive modeling-based knowledge acquisition for photo interpretation." Paper presented at the Intelligence Community's Sixth Annual AI Symposium, Crystal City and Langley, VA. Siegel, S. & Castellan, N.J. Jr., (1988) Nonparametric Statistics for the Behavioral Sciences (2nd Edition). New York: McGraw-Hill, Inc. SPSS [Statistical Package for Social Scientists], (2000) van der Horst, R. (1991) Time-to-collision as a cue for decision-making in braking. In: Gale, A.G., Brown, I.D., Haslegrave, C.M., Moorhead, I., & Taylor, S. (Editors) Vision in Vehicles—III. North-Holland: Elsevier Science Publishers, B.V., 19-26 Winer, B.J., Brown, D.R., & Michels, K.M. (1991) Statistical Principles in Experimental Design (3rd Edition). New York: McGraw-Hill, Inc. Detailed Design Report 138 August 10, 2001 5.0 vehDAQ23 DESIGN The main focus of the Field Operational Test is to improve the ability of a driver to safely navigate and guide a vehicle under conditions of low visibility. The implementation of this driver assistive system subjects the driver to a variety of new displays and stimuli. It is imperative to document that these new stimuli provide intuitive feedback to the driver, decreasing the stress and mental workload associated with the operation of the specialty vehicle. The most direct way to analyze the response of the driver to this new technology is to directly record the response of the driver to it. To record the response of the driver to these new stimuli, a variation of the “miniDAS” developed by NHTSA will be used both in the human factors driving simulator and the six test vehicles. This vehDAQ will record 4 separate video records (three cameras aimed at the driver, and one camera aimed forward out the vehicle windshield (as required by the original US DOT RFP) and one audio record. 5.1 BACKGROUND Variations of the microDAS data acquisition system have existed for a number of years. The original DAS was designed to record the response of typical drivers to situations which arise in everyday driving. The DAS would be installed in a vehicle, and configured to record the driver response under particular situations. For instance, the DAS might be configured to record the response of a vehicle and driver to an emergency braking. The DAS would be triggered by high brake pressure indicating severe braking; as soon as severe braking was detected, the DAS would record the response of the driver via video and audio and the response of the vehicle via accelerometers, gyros, and other instrumentation. This data would be recorded on a computer hard drive; these hard drives would be collected at regular intervals on the order of six months. Recently, a new variation of the DAS has been used to analyze the magnetic tape lateral guidance system. A compact version of the original DAS was developed; this version is triggered when the vehicle detects the presence of the magnetic tape. Once the tape is detected, the vehDAQ is activated, and audio, video, and engineering unit data is recorded on the system hard drives24. As the vehicle departs from the test site, the tape is no longer detected, and the vehDAQ no longer records driver and vehicle data. 23 In the TRS, this was referred to as a microDAS following the precedent set by NHTSA with miniDAS. However, the acronym DAS has been used to identify the Driver Assistive System. Furthermore, commercial products exist which are called microDAS, so as to avoid any possible copyright infringement, what was previously referred to as microDAS is now known as vehDAQ. 24 These hard drives are removable which facilitates convenient analysis away from the vehicle. Detailed Design Report 139 August 10, 2001 Audio 1 Camera 1 Camera 2 Video Quad Unit MPEG II Video/Audio Real Time Compression Titler To hard drive data storage Camera 3 Video Capture and JPEG compression board Camera 4 time stamp for synchronization engineering unit data VehDAQ computer Vehicle computer ethernet Figure 43. vehDAQ system functional block diagram For this field operational test, the vehDAQ will be configured similarly to this system. Triggering of the vehDAQ will be determined by the DGPS signal; when the vehicle is determined to be on the test corridor, the vehDAQ will be activated. Engineering unit data (i.e., vehicle location determined by DGPS and magnetic tape, acceleration from the IMU, etc.) will be provided from the vehicle main computer via an ethernet connection; this eliminates duplication of sensors and signal processing equipment. The functional block diagram of the vehDAQ system is provided in figure 43 above. 5.2 SYSTEM OVERVIEW The vehDAQ will utilize a minimum of four cameras; per the RFP, one of these cameras will be aimed forward out the windshield. (This camera view will also record data necessary to compute the Motorist Visibility Index.) This vehDAQ uses color cameras to view the driver’s face and hands; through the development process, color cameras were found to provide backlight compensation to a degree far exceeding that of B&W cameras. Detailed Design Report 140 August 10, 2001 Because of the wide ambient light conditions to which a specialty vehicle is exposed, high quality back light compensation is critical. The camera aimed at the driver’s feet is a B&W camera with integral infrared illuminator. B&W is adequate for recording the response of a driver’s feet because ambient light conditions remain fairly constant underneath the dashboard. These cameras are shown in pictures 44 and 45 below. B&W Camera with Integral IR illuminator Color Cameras with backlight compensation Picture 44: Cameras used for capturing the driver’s feet, hands, and face, respectively Picture 45: Forward looking, high resolution camera Detailed Design Report 141 August 10, 2001 Camera 1 320 x 240 320 x 240 320 x 240 320 x 240 Camera 2 Camera 3 Camera 4 Each camera 640x480 Image size after quad unit Quad Unit Figure 44. Functional block diagram of the quad unit showing how 4 separate images are compressed into a single image The camera recording the view out of the windshield is a much higher quality camera than those used to record driver behavior. This is due to the requirements of the MVI algorithm. Accurate MVI calculation requires an extremely high quality image; the quality of the driver camera The output of these cameras are sent to both a quad unit which takes 4 images at 640x480 resolution and creates a single composite image of the four images at a resolution of 640x480 pixels. The functional block diagram of this process is provided in figure Quad. Once the composite image is created, a time signature is written onto each frame by the titler. This time signature allows images to be synchronized with the engineering unit data also collected by the vehDAQ computer. The output of the titler and the output of the vehicle microphone are then routed to the MPEG II video/audio capture/compression board where the audio and video data are interlaced, compressed, and subsequently written to a removable hard disk. The MPEG II compression handles 30 frames per second from the titler and compressed that data at a rate of 1.5 Mbits/seconds. The 1.5 Mbit/s rate appears to be the optimum rate when image quality and data storage requirements are both considered. In parallel is the capture, compression, and storage of the view out of the windshield. A separate processing path is required because the MVI algorithm requires a full 640x480 image to be recorded one time per second. The second path takes the output of camera 4, captures and performed JPEG compression at 1 Hz, and writes those images to the same hard drive as the MPEG II video data. Detailed Design Report 142 August 10, 2001 The final data path is the synchronization and collection of engineering unit data. This engineering unit data is collected by the vehicle computers which support the driver assistive system. Data packets here are collected, time stamped, and sent to the vehDAQ computer for storage. The time stamp on these data packets are used to synchronize the video images captured by the vehDAQ. 5.3 HARDWARE SPECIFICS The vehDAQ is comprised of the following components as indicated in table 14 below: Peripheral Component Vendor/Model Forward Camera COHU/1322-1000/EH13 Driver color cameras Marshall Electronics/1/3” Color camera, V-1255-bnc B&W bullet camera Hosfelt Electronics/ 75-340 Mini Mono Quad Unit SuperCircuits/QS21 Video Titler host adapter Audio Microphone Decade Engineering/BOBII-NTSC SuperCircuits/ #PA3 Comments ½” color CCD, 12mm focal length, F1.4 lens, CS mount Color camera with integral backlight compensation Bullet camera with IR illuminator Mini image quad unit for color images Video titler used for video / engineering unit data synch Microphone and preamp Table 14. List of peripheral vehDAQ components Detailed Design Report 143 August 10, 2001 Computer Component Single board computer MicroBox Chassis Power supply Computer RAM SCSI controller card SCSI hard disk Vendor/Model Comments Cyber research / CPCD COP-800-C Cyber Research / MB IPC6AP Cyber Research / PSA 25112 Multiwave / PC133 256MB Adaptec / 29160N Quantum / QM336700TYLW Single board computer, 800 MHz Pentium III processor Removable HD frame StorCase / DE100 MPEG II Compression board DV Studio Technologies / Apollo Expert Plus JPEG compression board Videum AV by Winnov MicroChassis, 6 PCI slots 12V to 12V Switching power supply, 250 Watt 256 MB DIMM Ultra 160 controller Ultra 160 36 GB SCSI hard drive Removable hard drive and ancillaries Real time MPEG II encoding / decoding board Real time JPEG image compressor Table 15. vehDAQ computer components The assembled vehDAQ computer is shown in figures 46 - 48. Slot for removable hard drive Picture 46: Front of vehDAQ computer showing slot for removable hard drive. Detailed Design Report 144 August 10, 2001 Picture 47: Removable hard driver carrier with cover removed showing hard drive. Note lock on lower right corner. This prevents the unauthorized removal of video, audio, and engineering unit data. Ethernet Port MPEGII Compression board JPEG compression board port Picture 48: Business end of vehDAQ computer showing video, audio, and ethernet connections. Detailed Design Report 145 August 10, 2001 The vehDAQ has been designed to operate reliably in a harsh (in terms of temperature and vibrational exposure) environment. Industrial grade PC components were selected for both computer and peripheral components. One thing acting in favor of the vehDAQ is the fact that for all vehicles, the vehDAQ system is located inside the vehicle. It therefore takes advantage of all of the suspension and compliant body mounts which act to reduce the magnitude of the vibrations to which the vehDAQ is exposed. One key feature of the vehDAQ is the lockable, removable hard drive. Two hard drives are included with each vehDAQ computer; one IDE drive holds the operating system and the custom code written to execute its operation; the other hard drive is an Ultra 160 SCSI hard drive which is mounted on a removable frame/carrier. The bandwidth of the SCSI hard drive is required to accommodate the 1.5 Mbit/s data rate from the MPEG II compression. This ensures that data is passed without any loss. The removable hard driver is removed at the end of a shift, and is replaced with another. The freshly removed hard drive is then placed in another computer, and a DVD-ROM archive of the data on that hard drive is created. After the data is archived, the hard drive reenters service. Engineering unit, MPEG II video, and JPEG data required a storage rate of approximately one GB per hour. Assuming that the test vehicle operates on the test corridor, a full 12 hour shift will fit onto two DVD-ROM disks (assuming double sided disks at 4.7 GB per side). At least 1825 removable hard drives will be required to support the six vehicles under test. 5.4 SOFTWARE DETAILS All vehDAQ software was written in-house. Originally, the vehDAQ was to be purchased from Riverside Technical Design (RTD), Riverside, Iowa. However, RTD closed shop in Summer, 2000, leaving a gap in the program. Because these units are unavailable commercially, design and development was done by project partners. The vehDAQ runs under Microsoft Windows 2000 instead of the QNX Neutrino used by the other on-board vehicle computers. This is because drivers for the MPEG II and JPEG video capture/compression boards are unavailable. Software development kits for these boards are also only supported under the Windows operating system, further forcing development in that direction. The vehDAQ boots up, starts program execution, stops program execution, and shuts down based on commands to the Windows 2000 operating system from the vehicle control computers. The only tricky bit is the shutdown procedure. Unlike QNX Neutrino, Windows needs an orderly close of system files for a proper shutdown. An external timer is used to control the vehDAQ shutdown. When the Neutrino based vehicle computer 25 More might be required; the final number depends on the operations plan presently under development. Detailed Design Report 146 August 10, 2001 gives the command to the vehDAQ, a delay switch is also activated. The vehDAQ computer is given a specific time period in which to shut itself down. When that time period is complete, the power is cut from the vehDAQ computer. 5.5 ENGINEERING UNIT DATA ACQUSITION The following data is collected and stored at the 10 Hz rate: (1) (2) (3) (4) (5) (6) (7) (8) GPS Position (velocity derived from position) Vehicle orientation (heading, roll, pitch) Forward radar returns. Three axes acceleration Three axes rotational rates Steering position Brake apply Turn signal activity Detailed Design Report 147 August 10, 2001 Appendix A - Lane Awareness System Preliminary Product Bulletin for Snow Removal Operations Description 3MTM Lane Awareness System for snow removal operations helps to improve the personal safety of the snow removal operators, increases the level of service and helps to improve the safety of the traveling public, while lowering the cost of the snow removal operation and reducing the economic impact on the region. 3MTM Lane Awareness System for Snow Removal Operations consists of three matched components: 3MTM Sensing Electronics 3MTM Magnetic Tape Series 2000 3MTM Operator Display and Interface A rugged electronic package, which has been designed and tested to withstand the tough environments and temperatures, encountered during snow removal operations. It contains the sensing and signal processing electronics and connects to the vehicle to obtain power and speed information. It senses the presence of the magnetic tape, calculates its distance from the tape and transmits this information to the operator display. TM The 3M Magnetic Tape Series 2000 marks the pavement with a magnetic reference line, which is tracked by the sensing electronics. It is a preformed, patterned, skid resistant pavement marking tape with pressure-sensitive adhesive. The white or yellow magnetic tape can replace the existing standard solid long or skip lines. 3M requires grooved or underlay installations of all 3MTM Magnetic Tape Series 2000 when used for snow removal applications. The in-vehicle display continuously shows the operator how close the sensing electronics are from the magnetic reference tape. The operator can adjust the display, select the left or right sensing electronics (if more than one is installed on the vehicle), and set the warning distance. The operator interface warns the driver when the distance exceeds the set warning distance. The warning can be selected to be a visual warning and/or a vibrotactile warning provided through the vibration of the operator’s seat. System Features Resistance to Demagnetization: Intrinsic Coercivity (Hc) is 300 kA/m or higher. Magnetic Temperature Dependence: Temperature coefficient of Remanent Magnetization: < 500A/m/ºC: 3MTM Magnetic Tape Series 2000 Model 2000: Conformable white magnetic tape which meets standard highway colors and with a retroreflectivity of 500 millicandelas per square meter per lux (mcd/m2 lux) at an entrance angle of 86.5º and observation angle of 1.0º. Model 2000S: Continuous black magnetic tape with white custom skip lines. Model 2001: Conformable yellow magnetic tape which meets standard highway colors and with a retroreflectivity of 300 mcd/m2 lux at an entrance angle of 86.5º and observation angle of 1.0º. Model 2001S: Continuous black magnetic tape with yellow custom skip lines. Model 2005: Conformable black magnetic tape. Retroreflectivity: Durable retroreflectivity is achieved by abrasion-resistant microcrystalline ceramic beads bonded in highly durable polyurethane. Skid Resistance: 55 BPN when installed and not less then 45 BPN after four years when tested according to ASTM E303. Adhesive: Precoated with a pressure sensitive adhesive on bottom surface. Disposal Requirements: Manufactured without the use of heavy metals, lead chromate pigments or other similar, lead-containing chemicals and, at present, without special disposal requirements. Magnetic Pattern: Magnetic field reverses every meter. Detectable Magnetic Field: 0.3 Gauss minimum above background levels at 20º C (68º F) when measured at the center of the tape and 50 cm (20 in.) from the magnetic transition. Preliminary –40 C = -0 to +10% on nominal referenced to +20 C (68 F) +80 C = -10 to + 0 % on nominal referenced to +20 C (68 F) Tape Width: Ten centimeters (4 in.) standard. Nominal Thickness: 2.3 mm (0.090 in. ) at pattern heights Installation: Installed as underlay under the last lift of asphalt or grooved into surface with grooving depth of at least 3.0 mm (0.1 in.). Shipment: Shipped in cartons containing four rolls of 60 meters (196.8 ft.) using standard carriers (magnetic field less than 0.002 Gauss at 2.1 m (7 ft.). Storage: Store in a cool, dry indoor area. Use within one year of receipt. 3MTM Sensing Electronics Dimension: 61 cm long, 6 cm wide and 2.5 cm deep (23.8 in. by 2.4 in. by 1 in.). Operating Voltage: 12 Volts DC obtained from vehicle. Vehicle Connection (other than power): Connects to J1708 vehicle bus vehicle to obtain vehicle speed. Optimal Vehicle Speed: Optimal operation at vehicle speeds ranging from 8 to 80 kph (5-50 mph). Initial Signal Acquisition Distance: Operational after detecting the presence of the tape for four meters (13 ft.) when within detection range (distance from tape less than 90 cm (3 ft.). Calibration: self calibrating. 1 IF 5.3: Adhesive Spray Applicator (PS-14) IF 5.7: Pavement Surface Preparation and Application Techniques For 3M Stamark™ Tapes. Output: calculates distance from the center of the sensing electronics to the center of the tape. Detection Range: ±90 cm (±3.1 ft.) from the center of the magnetic tape (validated in laboratory tests). Distance Error: ±1 cm above the tape and ±5 cm at ±90 cm (±2 in. at ±3.1 ft.) from the center of the tape at speeds from 8-40 kph (5-25 mph) (validated in laboratory tests). Operational Height: 15cm to 45 cm (6 in. to 18 in.) above the tape with optimal performance at 20 cm (12 in.). Dynamic Height Changes: Detection range and accuracy are independent of dynamic changes in height. Distance Update Frequency: 0.5 sec at 16 kph (10 mph) and 0.1 seconds at 80 kph (50 mph). Operating Environment: –40º C to 80º C (-40º F to 176º F) and 10 to 99% humidity (non-condensing) over ambient temperature range. Self-Test: Sensing electronics performs a self-test at start up and failures are displayed on the in-vehicle display. RFI and EMC Requirements: Passed all applicable FCC requirements for RFI and European EMC requirements. IF 5.13 Instructions For Using3M Stamark™ Contact Cement E-44T IF 5.14: Manual Highway Tape Applicator (MHTA-16 and MHTA 18). IF 5.15: Motorized Manual Highway Tape Applicator (MMHTA-18). IF 5.17 Instructions for Using 3M Stamark™ Surface Preparation Adhesive P-50 for 3M Stamark™ and Scotch-Lane ™ Pavement Marking Tapes. Installation Guides Climate Guide for 3M Stamark Pavement Marking Tapes. Road Surface Guide for 3M Stamark Pavement Marking Tapes. Adhesion Guide for 3M Stamark Pavement Marking Tapes. Health and Safety Information 3MTM Operator Display and Interface Read all health hazard, precautionary, and first aid statements found in the Material Safety Data Sheet (MSDS), and /or product label of chemicals prior to handling or use. Also refer to the MSDS for information about the volatile organic compound (VOC) content of chemical products. Consult local regulations and authorities for possible restrictions on product VOC content and/or VOC emissions. Dimension: 21.6 cm by 15.9 cm and 5.4 cm deep (8.5 in. by 6.25 in. and 2.125 in. deep). Operating Voltage: 12 Volts DC obtained from vehicle. Sensing Electronics Interface: Standard EIA RS-485 interface. Display Resolution: The display shows distance from the magnetic tape with a resolution of 1 cm (0.39 in.). Display Update Frequency: 10 updates/sec. Distance Warning Settings: Two adjustable settings are available – one to the left and one to the right side of the magnetic tape. The settings are adjustable over the entire detection range. Operator Warning: The operator receives a warning when the distance of the sensing electronics from the center of the tape exceeds the distance warning setting. The system requires installation of either the peripheral visual warning lights or the vibro-tactile warning system. Peripheral Visual Warning Lights: These lights are located in the operator's peripheral field of vision and warn the driver by blinking the left or right warning lights. Vibro-tactile Warning: The vibro-tactile warning system warns the driver by vibrating the left or right side of the operator's seat. Out-of-Detection Indication: Arrow points in the direction where the tape was last detected. Operating Temperature:–40º C to +80º C ( -40º F to 176º F and 10 to 99% humidity (non-condensing) over ambient temperature range. Literature References Installation Instructions IF 5.19: Pavement Surface Preparation and Application Techniques for 3M Magnetic Tape Series 2000. Information Folders (IF) for Application Equipment, Surface Preparation Adhesives and Contact Cements IF 5.1: Manual Highway Tape Applicator (MHTA-1) Preliminary 2 Limited Warranty and Remedy Information The following warranties are made in lieu of all warranties express or implied, including any implied warranty of merchantability or fitness for a particular purpose. 3M Magnetic Tape Series 2000 Limited Warranty Exclusive Remedy Subject to the conditions stated below, 3M warrants that the 3M Magnetic Tape Series 2000, which is sold as a component of 3M’s Lane Awareness System for Snow Removal Operations, will meet its published specifications at the time of shipment and will meet its durability specifications from the date of installation for the period stated below when used under normal traffic conditions and when inspected and verified as described below. For a period of four years from the date of installation, the 3M Magnetic Tape Series 2000 tape will retain sufficient magnetic material to meet or exceed eighty percent (80%) of all specified magnetic requirements, The 3M Magnetic Tape Model 2000 (white) and 2001 (yellow) and the retroreflective portion of the 3M Magnetic Tape Model 2000S (white skip-line) and 2001S (yellow skip line) will meet the minimum retained coefficient of retroreflection value of 100 millicandelas per foot squared per foot-candle (measured at 1.0 degree observation and 86.5 degree entrance angles) for a period of four years from the date of installation. All of the 3M Magnetic Tape Series 2000 tapes will adhere to the roadway for a period of four years from the date of installation. If the 3M Sensing Electronics, the 3M Operator Display and Interface, or the 3M Magnetic Tape Series 2000 fail to meet specifications at the time of shipment; or if the 3M Sensing Electronics or the 3M Operator Display and Interface fail to continue to meet specifications for an additional two year period; or if any tape of the 3M Magnetic Tape Series 2000 fails to meet the stated warranty for the periods specified, then 3M’s sole obligation and purchaser’s and customer’s exclusive remedy shall be, at 3M’s option, for 3M either to replace or repair the product or material so as to restore the product or material to its original effectiveness or refund the purchase price. Replacement products will carry the remaining unexpired warranty of the products replaced provided that all conditions of this warranty are met in performing the replacement. Inspection and Verification Requirements To identify and verify a reflectance failure, a visual night inspection must be made with a 3M representative and a customer representative to identify areas of the installation which appear to be below the minimum retained reflectance values as defined in the warranty. Such areas shall be identified as “zones of reflectance measurement.” To qualify for the remedies available under this warranty, the zone of reflectance measurement must be at least 360 feet in road length and consist solely of either edge lines, center lines, or lane lines, but not in combination. To identify and verify a magnetic failure, inspection and measurement of magnetic strength must be made with a 3M representative and a customer representative to identify areas of the installation which appear to be below the minimum retained magnetic field strength as defined in the warranty. Such areas shall be identified as “zones of magnetic field measurements.” To qualify for the remedies available under this warranty, the zone of magnetic field measurement must be below the specified retention for a continuous two-meter length. Sensing electronics and Operator Interface and Display Limited Warranty Subject to the conditions stated below, 3M warrants that the 3M Sensing Electronics and the 3M Operator Display and Interface, which are sold as components of 3M’s Lane Awareness System for Snow Removal Operations, will meet their published specifications at the time of shipment and for a period of two years from the date of shipment. Warranty Conditions The above warranties are conditioned upon the following: All of the 3M Magnetic Tape Series 2000 tapes, the 3M Sensing Electronics, and the 3M Operator Display and Interface will be (1) used under normal traffic conditions; and (2) installed and/or applied in accordance with all 3M application and installation procedures provided in 3M’s product bulletins, information folders, and technical memoranda (all of which will be furnished to the applicator or installer upon request, see listing in this brochure); (3) purchaser and/or customer has maintained an accurate record of the dates of installation of tape products (upon which any durability warranty is based); (4) purchaser and/or customer has notified 3M of any alleged failure within thirty (30) days of the date upon which such failure was first observed by purchaser; and (5) the inspection and verification procedures stated below have been followed by purchaser and/or customer. The above warranties shall not apply in the event of (1) roadway substrate or pavement failure; (2) repair or modification by persons not authorized by 3M, (3) misuse, neglect or accident, or (4) damage by extreme atmospheric, weather or traffic conditions. Preliminary 3 Limitation of Liability EXCEPT AS PROVIDED ABOVE, 3M SHALL NOT BE FOR ANY INJURY, LOSS OR DAMAGE, WHETHER DIRECT, INDIRECT, INCIDENTIAL, SPECIAL OR CONSEQUENTIAL, ARISING OUT OF THE USE, MISUSE, OR THE INABILITY TO USE THE 3M LANE AWARENESS SYSTEM FOR SNOW REMOVAL OPERATIONS (OR ANY OF ITS COMPONENT PARTS), REGARDLESS OF THE LEGAL THEORY ASSERTED. NO STATEMENTS OR RECOMMENDATIONS NOT CONTAINED OR REFERENCED HEREIN SHALL HAVE ANY FORCE OR EFFECT UNLESS STATED IN AN AGREEMENT SIGNED BY AUTHORIZED REPRESENTATIVES OF 3M. THE REMEDIES SET FORTH HEREIN ARE EXCLUSIVE. 3M assumes no responsibility for any injury, loss or damage arising out of the use of a product that is not of 3M’s own manufacture. Where reference is made in the literature to a commercially available product, not made by 3M, it shall be the user’s responsibility to ascertain the precautionary measures for its use outlined by the manufacturer. For further information, contact: 3 Intelligent Transportation Systems Safety and Security Systems Division 3M Center Bldg. 225-4N-14 St. Paul, MN 55144-1000 3M Canada Company P.O. Box 5757 London, Ontario, Canada N6A 4T1 1-800-328-7098 1-800-224-2085 Fax 1-800-3M-HELPS 519-451-2500 651-575-5794 651-737-1055 Fax e-mail: [email protected] http://www.3M.com/its © 3M 2000 Preliminary 4 Appendix B - Lane Awareness System Pavement Surface Preparation and Application Techniques for 3M Magnetic Tape Series 2000 for Snow Removal Applications Information Folder 5.19 ITS May 2000 Table of Contents Introduction Magnetic Tapes Series 2000 Requirements for Snowplow and Mountainous Regions General Application Requirements Underlay Application of Long Lines: Edge, Lane or Channelization Lines Grooved Application of Long Lines: Edge, Lane or Channelization Lines Replacement of Magnetic Tape Segments Storage, Removal and Health and Safety Appendix A. Surface Moisture Test After Rain or in Marginal Weather Conditions Appendix B. Pavement Surface Types Appendix C: Magnetic Tape Definitions Literature Reference Important Notice to the Purchaser 1 1 2 2 3 3 5 6 6 7 8 9 9 Introduction The installer is responsible to contact the 3M sales representative or 3M technical service representative whenever there is a question regarding application procedures or conditions. A 3M representative is required to be present during the installation of the magnetic tape. Please call 1-800258-4610 or send an email to [email protected] to request a technical service representative This information folder contains pavement surface preparation requirements and application procedures for 3M™ Magnetic Tape Series 2000. It is important that users be completely knowledgeable of all application requirements and procedures prior to product application. Instructions contained in this folder must be followed for material replacement provisions to be considered valid. Material replacement provisions are described in the appropriate product bulletins. . Magnetic Tape Series 2000 The surface preparation and installation instructions are for the following magnetic tapes: Model Magnetic Tape Application 2000 White Lines Grooved 2000S White Skip Lines Grooved 2001 Yellow Lines Grooved Draft 2001S Yellow Skip Lines Grooved 2005 Black Lines Grooved Black Lines Underlay All series 2000 magnetic tapes have pressure sensitive adhesive. 1 Draft Requirements for Snowplow and Mountainous Regions General Application Requirements The magnetic tape has to be installed either by the underlay or grooved application process for all applications in snow removal or mountainous regions. 3M does not warrant an overlay or inlay application of magnetic tape in mountainous and/or snow removal areas. The Lane Awareness System requires an uninterrupted magnetic line for its performance. Therefore, 3M requires that the magnetic tape is applied as a full length marking. For skip line application, the continuous marking can be either a continuous black magnetic tape with the white or yellow skips spliced in during manufacturing or else a continuous black magnetic tape with standard white or yellow road marking paint applied in the required skip pattern. The following requirements apply to all magnetic tape applications: Temperatures Air Temperature: Minimum 60°F (16°C) and rising. Overnight Air Temperature: Minimum 40°F (5°C) the night before tape application. Surface Temperature: Minimum 70°F (21°C) and rising. Pavement Surface The pavement surface must be clean and dry. No rainfall should occur within 24 hours prior to application. (See Appendix A for Rain and Marginal Weather Conditions.) Tape Splices Butt splices must be used; do not overlap tape ends. Traffic Control Traffic must be kept off of pavement surfaces coated with a surface preparation adhesive prior to tape application. Draft 2 Draft Underlay Application of Long Lines: Edge, Lane or Channelization Lines NOTE: This is one of two possible application procedures. Magnetic tapes with pressure sensitive adhesive can be applied underneath fresh asphalt before the wear course has been paved. Underlaying is the process of installing the tape underneath the last lift of asphalt. Note: The magnetic tape locations have to be marked accurately since an underlay installation permanent. 2. Apply the magnetic tape using the 3M Manual Highway Tape Applicator on the base or intermediate course before the tack or prime coat is sprayed and the surface course is placed. Note: The magnetic tape has to be installed immediately prior to the applying the last lift of asphalt. If the final pavement is delayed by more than 6 hours, a pre-adhesive needs to be applied to insure that the magnetic tape remains in place. New Road Surface Final Asphalt Lift Magnetic Tape Figure 1 Underlay Installation: Magnetic Tape installed under the last layer of asphalt 3. Procedure 1. Apply the tape. Pre-mark the road. Tamp the tape. Use a 3M Roller Tamper Cart RTC-2 loaded with 200 lbs. to get initial tack of the magnetic tape to the base or intermediate course. Use 200-250 ft. (60-70 m) of light rope, chain, or thin wire cable to mark the specified location of the magnetic tape. The edge or centerline locations that will exist after the lanes are paved and marked, are the most likely locations for the magnetic tape. 4. Pave the lanes. The tack or prime coat is applied and the lanes are paved. Grooved Application of Long Lines: Edge, Lane or Channelization Line NOTE: This is one of two possible application procedures. Procedure In mountainous and snow removal areas, grooving the pavement surface allows the magnetic tape to be applied onto an existing pavement surface. Grooving enhances adhesion by creating a fresh surface and protects the tape from damage by snowplow blades. All weather and climate conditions for overlay applications must be met for grooving specifications (See Appendix A: Surface Moisture Test after Rain or in Marginal Weather Conditions) Draft 3 Draft 4. Pavement Grooving Specifications: The bottom of the groove should have a smooth/flat finished surface. If a coarse tooth pattern is present, increase the number of blades and decrease the number of spacers on the cutting head. 3M will not be held accountable for applications made in poor quality grooves. The following recommendations must be followed to qualify for any material replacement provisions. 1. Cutting Head - Gang stacked 5. 0.65-1.25 cm (1/4-1/2 inch) wide diamond tipped cutting blades are recommended to produce the best-finished grooved surface. The spacers between each blade must be such that there is less than a 0.3 mm (10 mil) raise in the finished groove between the blades. Water cooling the blades may be necessary for edge line grooving. Note: Grooving asphalt tape applications requires the use of new blades and watercooling. This will result in a smooth grooved surface and will prevent the formation of ridges in the groove. Allow drying for 24 hours between grooving and tape installations. 2. 6. 7. Groove cleaning If cooling the blades with water is necessary, flush the groove with water immediately after grooving to clean the surface. Allow drying for 24 hours prior to tape application. Tape Application Procedure Summary Follow the climate and weather recommendations listed in General Application Requirements. 1. Clean the groove. A high-pressure air blower is recommended for cleaning all debris from the grooved surface. The surface of the groove must be dry. (See Appendix A.) Depth of Groove Tape thickness + 10% Example: For magnetic tape series 2000, the required depth of the groove is at least 2.5 mm (tape thickness of 2.3 mm plus 0.23 mm or 10%) or at least 100 mils (tape thickness of 90 mils plus 9 mils or 10%). 2. Apply the adhesive. Using a 3M™ Spray Applicator PS-14 (or a compressed air sprayer), apply a coat of P-50 adhesive to the pavement. The adhesive should cover the entire grooved surface. The PS-14 applicator is designed to spray a 15.24 cm (6inch) wide pattern for application of 10 cm (4 inch) wide tape. Note: One coat of P-50 sprayed at a 15-cm (6-inch) width for a 10-cm (4-inch) line covers approximately 36 lineal meters/liter or 450 lineal ft/gallon. Alternate surface preparation adhesives are available, please contact an Intelligent Transportation Systems representative for further information. Adjust the arm of the PS-14 applicator up or down so that the spray pattern covers the entire grooved surface (See Adhesive Spray Applicator PS-14, Information Folder 5.3 for detailed instructions). Figure 2 Schematic of Grooved Magnetic Tape Road Surface Magnetic Tape Width of Groove Installation 3. Draft Groove position The recommended position of the groove is a minimum of 2.5 cm (2 inches) from the edge of a longitudinal seam. Width of Groove Depth of Groove Grooving speed Speed will vary with width of the groove, size of application, and equipment being used. Tape width of 10 cm plus 2.5 cm +/- 3.2 mm or 4 inches plus 1 inch +/- 1/8 inch. 3. Finished surface 4 Allow the adhesive to dry completely. Draft Normal drying time is approximately 5 minutes at 10°C (50°F). It is very important that P-50 adhesive is dry before the tape is applied. If the from the crack or joint on each side (See Figure 3). 5. Cut Away Magnetic Tape Magnetic Tape ! ! Expansion Joint Pavement Crack ! 6. Tamp the tape. VERY IMPORTANT! Tamp the tape thoroughly with the 3M RTC-2 tamper cart with a minimum 200-pound (90-kb) load. Tamping is most important! Do not twist or turn the tamper cart on the tape. Make six passes (three passes back and forth) over each part of the tape. Make sure all edges are firmly adhered to the road surface. Open the road to traffic. As soon as you are finished tamping the tape on the pavement, you may open the road to traffic. textTape Magnetic text Special Tape Installation Requirements for Intersection Installations Note: This section applies to all magnetic tape installations that will experience cross traffic. It applies especially to those installations with vehicles turning while driving over the magnetic tape. 3M's warranty requires that E-44T contact cement be used when tape is installed at locations that experiences cross traffic or vehicles turning over the tape. E-44T can be applied to all grooved road surfaces except for new, less than 11 day-old asphalt. 1. Apply a uniform coat of E-44T using a long, (2-cm or 0.5 inch) nap paint roller with a solvent-resistant core. 2. Allow the E-44T to dry completely (see Step 3, Tape Application Procedure Summary). 3. Continue with Step 4, Tape Application Procedure Summary) P-50 adhesive is not allowed to dry, it will not bond properly to the adhesive on the tape and adhesion failure will likely occur. P-50 adhesive is dry when it will not lift or string when touched with fingertips protected with gloves. Figure 3: Illustration of the 2.5 cm or one inch cut back at expansion joints or pavement cracks. 4. Apply the tape. Apply the tape using the 3M Manual Highway Tape Applicator (MHTA) or the 3M Highway Tape Applicator (HTA). Note: If there is a crack in the pavement or if the tape is to be applied over an expansion joint, lay the tape over the crack or joint, then cut the tape 2.5 cm or one inch away Replacement of Magnetic Tape Segments magnetic field: the red colored end points in the direction of the magnetic north pole. This procedure explains how to replace magnetic tape on a concrete or asphalt. Before you begin, make sure your environment meets all the product General Application Requirements on page 2. For information or assistance: contact Intelligent Transportation Systems Project of 3M Safety and Security Systems Division at 1-800-258-4610 in the U.S. Locating the Magnetic Field Reversal The magnetic field reverses from magnetic field direction north to magnetic field direction south every meter producing a pattern that repeats itself every two meters. This produces a wavelength of two meters. The Magnaprobe will show the direction of the magnetic field when placed horizontally over the magnetic tape: the red magnet end points towards the sky when the magnetic field direction is north and towards the road when the magnetic field direction is south. The location Identifying the Magnetic Field Direction The magnetization pattern of the tape can be found by using a Magnaprobe Mark II. The Magnaprobe indicates the magnetic field by a small magnet located on the end of the probe. The magnetic aligns itself with the direction of the Draft 5 Draft ! ! ! of magnetic field reversal is identified when the red and blue magnet ends flip direction and are parallel to the tape surface. Replacement Procedure 1) Determine the amount of magnetic tape that needs to be replaced in two-meter lengths. the old tapes are well worn, if they tightly adhere to the road surface and if the grooving is deep enough to meet the specification for the grooving depth. 3) Clean the road. Clean the surface of the road before the magnetic tape is applied. A high-pressure air blower is recommended for cleaning all debris from the groove surface. The surface of the groove must be dry. (See Appendix A.) The amount of tape to be replaced depends on the damaged region and the magnetization pattern. The magnetic field reverses every meter (from north-south to south-north). This pattern repeats itself every two meters producing a twometer wavelength. Therefore, magnetic tape has to be replaced in increments of two meters or one wavelength. Locate at least two magnetic field reversals beyond the damaged region and mark them on the magnetic tape (See Locating the Magnetic Field Reversal above). 4) Tape Application Procedure Summary Follow the application instructions for tape installation in grooved surfaces, as well as the climate and weather recommendations on page 2. For replacements where vehicles are making high shear or crossover turns, use Surface Preparation Adhesive E-44T as the application adhesive before magnetic tape application 2) Determine road surface conditions. If any of the old markings from the roadway surface by sandblasting or grinding it away. A minimum of 90% of the road surface under the existing markings must be exposed prior to tape application. The new magnetic tape may be applied over existing Stamark or Magnetic tapes only if Storage, Removal and Health and Safety product label of chemicals prior to handling or use. Also refer to the MSDS for information about the volatile organic compound (VOC) content of chemical products. Consult local regulations and authorities for possible restrictions on product VOC content and/or VOC emission. Storage Store away from heat in a cool, dry place and protect from freezing. Store out of direct sunlight. Keep container closed when not in use. Keep container in well ventilated area. Contents may be under pressure; open carefully. Caution • Removal Gloves should be worn when using any magnetic tapes to prevent injury to hands. • Do not use a flame or torch to remove magnetic tapes. Burning may violate local, state or federal air quality regulations. Also, exposing the tape to very high temperatures with the use of a flame or torch could generate emissions that may be harmful to skin, eyes and respiratory system. Magnetic tapes are designed for permanent, non-removal applications. Contact your 3M representative for guidance. Health and Safety Read all health hazard, precautionary and first-aid statements found in the Material Safety Data Sheet (MSDS) and/or Appendix A. Surface Moisture Test After Rain or in Marginal Weather Conditions to overlay application. If you still question the dryness of the pavement after 24 hours following a rainfall, then perform the surface moisture test as follows: Magnetic tapes cannot be applied to a wet or damp surface. Rainfall before or during an application can prevent the tape from sticking to the pavement. Road surfaces must be dry and above the minimum required temperature for application of magnetic tapes. No rainfall should occur within 24 hours prior Draft 6 Draft Materials needed Figure A-1: Tape down a black (dark) plastic sheet over the application area using duck tape or magnetic tape. Polyethylene sheets approximately 18 in. x 18 in. square (45 cm x 45 cm square). A standard black or green garbage bag can be used. ! 5-cm. (2-in) wide duct tape or 10-cm. (4-in) wide Magnetic tape ! Utility knife ! Tamper cart 3. Firmly tamp the tape by using your foot or a tamping cart. Procedure 4. This procedure tells you how to determine if the road surface is dry enough to apply Magnetic tape. This test requires full sunlight. 1. Place the polyethylene sheet on the pavement. Wait 20–25 minutes. Allow 20–25 minutes for the sheet to be exposed to direct sunlight. 5. Remove the sheet. Remove the sheet from the road surface. If no moisture (condensation) is present on the backside of the sheet or on the road surface, Magnetic tape can be applied. In an area of direct sunlight, place one layer of polyethylene (garbage bag) on the pavement where the tape application will be done. Make sure there are not any holes or tears in the polyethylene sheet. The pavement should be dry and clean. 2. Tamp the tape. If moisture is present, allow another hour to pass and repeat the test until no moisture is found. Experience has shown that if moisture is present in the pavement, the tape will not adhere to the surface. Tape the edges of the sheet. Using the duct tape or pavement marking tape, tape all the edges of the sheet. The tape should seal all the edges and not allow any air movement to get under the sheet. (See Figure A1). Appendix B. Pavement Surface Types The pavement surface type and age determine which application procedures should be used for magnetic tapes. For pavement types not covered, contact your 3M representative. Open Grade Friction Course Asphaltic Cement Concrete (ACC) An open grade consists of only large aggregate, up to 3/4 inches (19 mm), with an asphalt cement binder. Generally, Magnetic tapes should only underlayed. Standard Mix Asphalt Chip Seal A chip seal is achieved by spraying a hot asphalt emulsion onto an existing surface and then covering it with a 1/4-inch (6.35-mm) aggregate. Excess aggregate is removed after several days of exposure to traffic. Magnetic tapes applied prior to chip seal application. Contact an Intelligent A standard mix contains both fine (sand) and large (1/4-inch [6-mm]) aggregate with an asphalt cement binder. Magnetic tapes can be underlayed or grooved. Draft 7 Draft Transportation Systems representative for specific instruction how to apply the magnetic tape prior to chip seal application. Portland Cement Concrete (PCC) New Concrete Slurry Seal The magnetic tape has to be installed using a grooving process. The groove has to be allowed to dry and cleaned before the surface preparation adhesive and tape are applied A slurry seal is a hot asphalt emulsion that is sprayed onto an existing surface, as a top coat dressing only. No aggregate is added. Magnetic tape Model 2005 can be installed by grooving the surface prior to applying the slurry seal. The colored magnetic tapes including skip lines can be installed by grooving the road surface after the slurry seal is applied. Old Concrete Old concrete is a Portland Cement Concrete surface that has been open to traffic for more than one year and is showing signs of wear. Often this surface is smooth with large amounts of exposed, polished aggregate. The magnetic tape has to be installed using a grooving process. The groove has to be allowed to dry and cleaned before the surface preparation adhesive and tape are applied Recycled Asphalt Recycled asphalt consists of asphalt that was removed, reprocessed and reapplied. Magnetic tape can be underlayed or grooved into the new pavement surface. Rubberized Asphalt This asphalt contains additives designed to increase its durability. It can be either a “standard mix” or an “open grade friction course.” Magnetic tape can be underlayed or grooved into the new surface. Appendix C: Magnetic Tape Definitions magnetic field direction north to magnetic field direction south every meter producing a pattern that repeats itself every two meters. This produces a wavelength of two meters, which repeats along the length of the tape. Magnetic Tape Refers to 3M Magnetic Tape Series 2000. It is produced by modifying the compound formulation of the 3M Stamark 380 Series Pavement Marking Tape manufactured by 3M Traffic Control Materials Division. The process replaces the tape filler with a magnetic pigment to achieve the desired magnetic properties. Wavelength Refers to the length of one alternating polarity pattern. The wavelength of the magnetic tape is two meters: magnetization to the north for one meter, then south for the next meter. Alternating Polarity Refers to the alternating magnetization of the magnetic tape. The magnetic field reverses from Draft 8 Draft Literature Reference For additional information on 3M™ Stamark™ Pavement Marking Tapes, application recommendations, or 3M application equipment, refer to the following publications: Product Bulletin 380I - 3M Stamark High Performance Tape Series 380I Information Folder 5.1 - 3M Manual Highway Tape Applicator (MHTA-1) Information Folder 5.2 - 3M Highway Tape Applicator (HTA) Information Folder 5.3 - 3M Adhesive Spray Applicator PS-14 Information Folder 5.4 - 3M Stamark Sprayable Adhesive SP-44T Information Folder 5.10 - 3M™ Fastbond™ Neutral Contact Adhesive NF-30 Information Folder 5.14 - 3M Manual Highway Tape Applicator (MHTA-16 and MHTA-18) Information Folder 5.15 - 3M Motorized Manual Highway Tape Applicator (MMHTA-18) 3M Climate Guide for 3M Stamark Pavement Marking Tapes 3M Road Surface Guide for 3M Stamark Pavement Marking Tapes 3M Adhesion Guide for 3M Stamark Pavement Marking Tapes Important Notice to the Purchaser 3M assumes no responsibility for any injury, loss or damage arising out of the use of a product that is not of our manufacture. Where reverence is made in literature to a commercially available product, made by another manufacturer, it shall be the user's responsibility to ascertain the precautionary measures for its use outlined by the manufacturer. All statements, technical information and recommendations contained herein are based on tests we believe to be reliable, but the accuracy or completeness thereof is not guaranteed, and the following is made in lieu of all warranties, express or implied. Seller's and manufacturer's only obligation shall be to replace such quantity of the product proved to be defective. Neither seller nor manufacturer shall be liable for any injury, loss or damage, direct or consequential, arising out of the use of or the inability to use the product. Before using, user shall determine the suitability of the product for his/her intended use, and user assumes all risk and liability whatsoever in connection therewith. Statements or recommendations not contained herein shall have no force or effect unless in an agreement signed by officers of seller and manufacturer. For information or assistance: 3 Intelligent Transportation Systems 3M Safety and Security Systems Division 3M Center, Bldg. 225-4N-1 St. Paul, Minnesota 55144-1000 1-800-328-7098 651-575-5494 651-737-1055 website: www.3M.com/its email: [email protected] 3M Canada P.O. Box 5757 London Ontario, Canada N6A4T1 519-451-2500 1-800-3MHELPS  3M 2000 Printed in U.S.A. 75-0500-4234-2 Draft 9 Draft Appendix C Communication Interface Standard for 3M Lane Awareness System 78-8114-5971-4 Rev 0.93 May 10, 2000 Prepared by: SEMS LAB, 3M 3M Center, Building 235-3F-08 St. Paul, MN 55144 Copyright © 2000, 3M Table of Contents 1. Introduction ................................................................................................................. 1 1.1 Scope ................................................................................................................. 1 1.2 Overview ........................................................................................................... 1 2. System Description...................................................................................................... 1 2.1 3M Lane Awareness System ............................................................................. 1 2.1.1 3M Magnetic Tape ............................................................................. 2 2.1.2 Sensor System .................................................................................... 2 2.2 Operator Interface System................................................................................. 2 3. Communication Interface Specification.................................................................... 3 3.1 Physical Network .............................................................................................. 3 3.2 Message Format ................................................................................................ 4 3.3 Timing ............................................................................................................. 10 3.4 Transmission Configuration............................................................................ 11 3.4.1 Baud Rate ......................................................................................... 11 3.4.2 Data Format...................................................................................... 11 3.4.3 Byte Format...................................................................................... 11 3.5 Error Detection................................................................................................ 11 3.6 Example........................................................................................................... 11 3M CONFIDENTIAL @2000 BY 3M CO. i PRINTED: 2/12/2002 COMMUNICATION INTERFACE STANDARD 3M LATERAL GUIDANCE TAPE & SENSOR SYSTEM 1. Introduction This document describes the communication interface to the 3M Lane Awareness System. 1.1 Scope This standard is intended to address all aspects of the communication interface between the Sensor System and an Operator Interface System. This communication interface standard serves as a binding information among designers, programmers, customers, and testers. It provides them with an agreement needed before proceeding with the detailed design of the design components. In addition, this standard may be used directly by customers in the production of a unique application specific Operator Interface Module. 1.2 Overview Section 1 provides an overview of the standard. Section 2 describes the 3M Lane Awareness System and the Operator Interface System. Section 3 describes the communication interface in detail. 2. System Description This communication standard will describe the interface between two systems: the 3M Lateral Guidance Tape & Sensor System and the Operator Interface System. 2.1 3M Lane Awareness System The 3M Lane Awareness System is composed of two components: 3M magnetic tape and a sensor system as shown below. PRINTED: 2/12/2002 1 COMMUNICATION INTERFACE STANDARD 3M LATERAL GUIDANCE TAPE & SENSOR SYSTEM Sensor System Operator Interface (Tape Detection and Distance Calculation System) (inside the vehicle) 3M Magnetic Tape (inlaid or underlaid in the road) Figure 1.0 3M Lateral Guidance Tape and Sensor System Sensor System SENSOR BOARD Sensors Filter Amplifier CONVERTER A/D converter PROCESSOR uP/Dsp Com Port Figure 2.0 Sensor System 2.1.1 3M Magnetic Tape The 3M magnetic tape is inlaid or underlaid the road’s surface. The tape can be placed at either the center or edge of a driving lane. 2.1.2 Sensor System The sensor system is placed along the front of the vehicle. The sensor system serves two purposes; first it detects or senses the magnetic tape on the road and then it uses this information to calculate the lateral distance away from the tape. 2.2 Operator Interface System The Operator Interface Display shall be placed inside the passenger cabin of the vehicle using the 3M Sensor System. The Operator Interface will be application specific. PRINTED: 2/12/2002 2 COMMUNICATION INTERFACE STANDARD 3M LATERAL GUIDANCE TAPE & SENSOR SYSTEM Depending on the vehicle used and its internal environment, the type of Operator Interface will vary. For example the Operator Interface could be visual, audio, physical effects, or a combination of any of those. The Operator Interface System will use the parameters passed through the communication interface plus any additional input it needs to process the desired information it will provide to the operator. The eventual output to the operator may be in any of the forms listed above or another. 3. Communication Interface Specification 3.1 Physical Network The system must have good noise immunity. The wiring will be exposed to severe vehicular environmental conditions including temperature extremes and engine compartment noise. Good shielding and a balanced transmission scheme is therefore necessary. A RS485 network meets these requirements. The Sensor System has a dedicated RS485 line to communicate with the Operator Interface. It uses a bi-directional implementation. This will allow ease of operation, a simple protocol, and will accomplish the desired transmission of information. Sensor System Distance Calculation Operator Interface System RS485 Operator Interface Figure 3.0 Physical Network PRINTED: 2/12/2002 3 COMMUNICATION INTERFACE STANDARD 3M LATERAL GUIDANCE TAPE & SENSOR SYSTEM 3.2 Message Format The Output messages from the Sensor System are 11 bytes and the Input messages to the Sensor System are 10 bytes. They have fixed length, with the format illustrated in the following tables. Table C1: Output Message from the Sensor System Format. Byte 1 2-3 4 Parameter STX Sequence Number Message ID Description Start Byte - 5-9 10 Data Checksum 11 ETX Lower 8 bits of the summation End Byte Range/Value < 0-99 0-9, A-Z (upper case characters) 0-FF Unit decimal 0: self-check 1: distance others: command Binary code > - Table C2: Input Message to the Sensor System Format. Byte 1 2-3 Description Start Byte The Sequence number of received message - Range/Value < 0-99 4 Parameter STX Sequence Number Message ID 5 Handshake Signal - 0-1 6-8 9 Data Checksum 0-999 0-FF 10 ETX Data Package Lower 8 bits of the summation End Byte PRINTED: 2/12/2002 a-z (lower case characters) > Unit decimal c: request self-check d: disable e: enable h: handshake r: reset t: test-mode v: velocity others: command 0: message error 1: message correct Binary code - 4 COMMUNICATION INTERFACE STANDARD 3M LATERAL GUIDANCE TAPE & SENSOR SYSTEM Field specification: • All fields except Checksum and Flag fields are ASCII codes. • The Sequence Number is increased by one per message sent by the Sensor System. It starts from 0 and increases to 99, then repeats the sequence starting at 0. The Sequence Number of the input message from the Operator Interface System is the Sequence Number of the most recent received message; • The Message ID field determines the message types, and the definition of each field is varied from different types of messages: Output Message from the Sensor System: • The Self-check Message provides the result of the Sensor System self-check to the Operator Interface; • The Distance Message contains the distance information from the Sensor System. This is the most frequently used message type in the interface; Input Message to the Sensor System: • The Request Self-check Message requires the Sensor System to perform self-check and send back the result in the Self-check Message format; • The Enable/Disable Messages work as a pair to put the Sensor System to idle state and wake it up. It can be used in the situation that is not suitable to calculate distance such as the snowplow is moving the plow up and down. • The Handshake Message informs the Sensor System if the last message has been correctly received; • The Reset Message is to reset the Sensor System. After receiving a reset, the Distance Calculation Module of the Sensor System will be re-initialized and recalibrated. The Sequence Number of the Distance Message will start from 0 again; • The Test-mode Message is to turn on or off the Sensor System test mode.; • The Velocity Message is to send the vehicle speed information to the Sensor System. • All the unused fields are filled with 0. For example, Distance fields are filled in 08 to represent that the distance is 8cm. • All the reserved fields are filled with 0; • The checksum is the lower 8 bits of the summation of the previous fields' binary value. For example, the checksum of <00101009 is 0x0c7. PRINTED: 2/12/2002 5 COMMUNICATION INTERFACE STANDARD 3M LATERAL GUIDANCE TAPE & SENSOR SYSTEM For Output Messages from the Sensor System, there are two message types defined so far: 1. Self-check message Table C3: Self-check Message Format. Byte 1 2-3 4 5-6 7-8 9 Parameter STX Sequence Number Message ID Reserved Reserved Status Description Start Byte Self-checked system status Range/Value < 0-99 0 00 00 0-FF 10 Checksum 0-FF 11 ETX Lower 8 bits of the summation End Byte Unit decimal 0x0: passed 0x80: in progress others: failed Binary code > - Note: After the sensor system received the self-check command from the operator interface system, it will start the self-check task. It will continue to send out message every 100ms, with the data as 0x80 to indicate that the self-check is in progress. When the task is done, it will send out a message include the code for the result. 2. Distance message Table C4: Distance Message Format. Byte 1 2-3 4 5-6 Parameter STX Sequence Number Message ID Distance Range/Value < 0-99 1 0-FF Flag Description Start Byte Distance related to the tape Flag Byte 7 8 SOT Side of Tape 0-2 9 Q Quality Factor 0-9 10 Checksum Lower 8 bits of the summation 0-FF PRINTED: 2/12/2002 0-FF Unit decimal cm (hexadecimal) 000000x1b: old distance 000000x0b: new distance 0000000xb: normal mode 0000001xb: test mode 0: right 1: left 2: unknown 0: poor 9: good other: medium Binary code 6 COMMUNICATION INTERFACE STANDARD 3M LATERAL GUIDANCE TAPE & SENSOR SYSTEM Byte 11 Parameter ETX Description End Byte Range/Value > Unit - Note: The Flag byte is binary code. The last three bits are used to indicate the distance information is an updated one or the same as previous, and to indicate if the Sensor System is in test mode. For Input Messages to the Sensor System, there are seven types of messages so far: 1. Request Self-check message Table C5: Request Self-check Message Format. Byte 1 2-3 4 5 6-8 9 Parameter STX Sequence Number Message ID Reserved Reserved Checksum 10 ETX Description Start Byte Lower 8 bits of the summation End Byte Range/Value < 00 c 0 000 0-FF Unit Binary code > - 2. Enable message Table C6: Enable Message Format. Byte 1 2-3 4 5 6-8 9 Parameter STX Sequence Number Message ID Reserved Reserved Checksum 10 ETX PRINTED: 2/12/2002 Description Start Byte Lower 8 bits of the summation End Byte Range/Value < 00 e 0 000 0-FF Unit Binary code > - 7 COMMUNICATION INTERFACE STANDARD 3M LATERAL GUIDANCE TAPE & SENSOR SYSTEM 3. Disable message Table C7: Disable Message Format. Byte 1 2-3 4 5 Parameter STX Sequence Number Message ID Handshake Signal Description Start Byte - Range/Value < 00 d 0-1 6-8 9 Reserved Checksum 000 0-FF 10 ETX Lower 8 bits of the summation End Byte > Unit 0: message error 1: message correct Binary code - Note: The Disable message informs the Sensor System to get into an idle state. The Sensor System stops sending out messages until it receives an Enable message. In another word, in idle state, the Sensor System does not transmit any message but still is ready to receive messages while it will ignore any message except Enable message. 4. Handshake message Table C8: Handshake Message Format. Byte 1 2-3 Parameter STX Sequence Number 4 5 Message ID Handshake Signal 6-8 9 Reserved Checksum 10 ETX PRINTED: 2/12/2002 Description Start Byte The Sequence Number of the most recent received message - Range/Value < 0-99 Lower 8 bits of the summation End Byte 000 0-FF h 0 or1 > Unit decimal 0: message error 1: message correct Binary code - 8 COMMUNICATION INTERFACE STANDARD 3M LATERAL GUIDANCE TAPE & SENSOR SYSTEM 5. Reset message Table C9: Reset Message Format. Byte 1 2-3 4 5 Parameter STX Sequence Number Message ID Reserved Description Start Byte - Range/Value < 00 r 0 Unit - 6-8 9 Reserved Checksum 000 0-FF Binary code 10 ETX Lower 8 bits of the summation End Byte > - 6. Test-mode message Table C10: Test-mode Message Format. Byte 1 2-3 4 5 Parameter STX Sequence Number Message ID Reserved Description Start Byte - Range/Value < 00 t 0 6-8 Data - 0 or 1 9 Checksum 0-FF 10 ETX Lower 8 bits of the summation End Byte PRINTED: 2/12/2002 > Unit 0: turn off test mode 1: turn on test mode Binary code - 9 COMMUNICATION INTERFACE STANDARD 3M LATERAL GUIDANCE TAPE & SENSOR SYSTEM 7. Velocity message Table C11: Velocity Message Format. Byte 1 2-3 Parameter STX Sequence Number Range/Value < 0-99 Message ID Handshake Signal Description Start Byte The Sequence Number of the most recent received message - 4 5 6-8 V Vehicle Velocity 0-999 9 Checksum 0-FF 10 ETX Lower 8 bits of the summation End Byte 0: message error 1: message correct kilometer per hour (decimal) Binary code > - 3.3 v 0 or 1 Unit decimal Timing The Sensor System shall use the message format listed above transmit to its data every 100ms. Between two transmits, the Sensor System will monitor and receive any data input from the Operator Interface System. The Operator Interface System is supposed to transmit message after receiving message from the Sensor System and finish transmission within the 100ms interval, as shown in Figure 4.0. Sensor Output message 0 Sensor Input message Sensor Output message 100 Sensor Input message Sensor Output message 200 Sensor Input message t(ms) Figure 4.0 Timing of the Communication Interface PRINTED: 2/12/2002 10 COMMUNICATION INTERFACE STANDARD 3M LATERAL GUIDANCE TAPE & SENSOR SYSTEM 3.4 Transmission Configuration 3.4.1 Baud Rate The baud rate used in the Sensor System is 9600. 3.4.2 Data Format The data is in NRZ (nonreturn-to-zero) format, which is defined as: one start bit eight data bits no parity bit one stop bit 3.4.3 Byte Format The data will be sent in Big Endian (MSB first). 3.5 Error Detection The Operator Interface will use the check sum for error detection and if it detects an error it should disregard that packet and wait for the next one. It may also use the sequence number of the received message as an index to check if there is any message missing. The Operator Interface may return the sequence number of the received message as a handshake signal. 3.6 Example The Sensor System detects the vehicle is 47cm right to the tape. It will send out the Distance Message to the Operator Interface System, assuming the sequence number is 5 and the Quality Factor is 9: <-0-5-1-2-F-0x0-0-9-0x0e3->. The Operator Interface System receives this message correctly according to the checksum. Current vehicle speed is 30 km/h. A Velocity Message will be received by the Sensor System: <-0-5-v-1-0-3-0-0x0db->. PRINTED: 2/12/2002 11 COMMUNICATION INTERFACE STANDARD 3M LATERAL GUIDANCE TAPE & SENSOR SYSTEM Revision Record Rev. Draft Draft Draft Draft Draft Draft Draft Rev 0.8 Rev 0.9 Rev 0.91 Rev 0.92 Page Rev 0.93 PRINTED: 2/12/2002 Paragraph Description Define a Message pair for the Operator Interface to request the Sensor System to do self-check and to get the result of the self-check from the Sensor System. Appvd Date 7/8/1998 8/26/1998 8/28/1998 9/15/1998 11/5/1998 2/9/1999 3/24/1999 6/3/1999 1/26/2000 2/28/2000 4/3/2000 5/10/2000 12 Installation Instructions Lane Awareness System Vehicle System Series 2000VS 3M™ Vehicle System Series 2000VS Installation i Table of Contents 1 About This Manual ..................................................................................................................................1 1.1 Purpose of Manual...............................................................................................................................1 1.2 Manual Conventions .............................................................................................................................1 1.3 Manual Organization.............................................................................................................................1 2 Safety Information ...................................................................................................................................2 2.1 2.2 2.3 2.4 Intended Use Statement........................................................................................................................2 Technical Support.................................................................................................................................2 Safety Messages..................................................................................................................................2 Safety Considerations ...........................................................................................................................3 2.4.1 Personal Safety Equipment and Clothing......................................................................................3 2.4.2 Electric Shock ...........................................................................................................................3 2.4.3 Explosion...................................................................................................................................3 2.4.4 Chemical Burns .........................................................................................................................3 2.5 Disposal of Product..............................................................................................................................3 3 Description ...............................................................................................................................................4 3.1 Lane Awareness System......................................................................................................................4 3.2 Vehicle System Series 2000VS .............................................................................................................4 4 Installation................................................................................................................................................7 4.1 Pre-Installation Vehicle Inspection.........................................................................................................7 4.2 Installation Procedures..........................................................................................................................8 4.2.1 Operator Display Mount.............................................................................................................9 4.2.1.1 Display Floor Mounting Bracket Installation .................................................................10 4.2.1.2 Display Dash Mounting Bracket Installation .................................................................11 4.2.1.3 Display Windshield Mounting Bracket Installation.........................................................12 4.2.2 Peripheral Vision Lights Installation...........................................................................................13 4.2.3 Junction Box Installation...........................................................................................................14 4.2.4 Junction Box Cable Connections ...............................................................................................15 4.2.5 Sensor Mounting Bracket Installation Guidelines ........................................................................16 4.2.6 Standard Sensor Mounting Bracket Installation ..........................................................................17 4.2.7 Sensor Bar and Cable Installation..............................................................................................18 4.2.8 Operator Display Installation.....................................................................................................19 4.2.9 Speed Input Cable Installation...................................................................................................20 4.2.10 Cable Conduit Installation .........................................................................................................21 4.2.11 Truck Power Connections ........................................................................................................22 5 Checkout.................................................................................................................................................23 5.1 Installation Checkout ..........................................................................................................................23 5.2 Performance Test ..............................................................................................................................25 6 Maintenance and Troubleshooting........................................................................................................26 6.1 Off-Season Component Storage ..........................................................................................................26 6.1.1 Sensor Bar Storage ..................................................................................................................26 6.1.2 Operator Display Storage .........................................................................................................26 6.2 Troubleshooting..................................................................................................................................27 3M™ Vehicle System Series 2000VS Installation 1 1.1 About This Manual 1.3 Manual Organization This manual is divided into six sections. Purpose of Manual This manual provides step-by-step instructions for installing 3M Vehicle System Series 2000VS components. It is intended for use by installers and maintenance personnel who are responsible for the installation and maintenance of the system. 1.2 1 Manual Conventions The conventions listed in Table 1-1 help to make this manual easier to use by presenting a uniform approach to the descriptions, phrases, and nomenclature. Section 1. About This Manual Contains information about the organization and content of this manual. Section 2. Safety Information Contains important information about safety messages and safety considerations for installation of this product. Section 3. Description Briefly describes the vehicle system sensing devices and operator interface components. Section 4. Installation Contains step-by-step installation instructions. Section 5. Checkout Contains information on how to set up and test the operator interface components and sensing devices. Section 6. Maintenance and Troubleshooting Contains information and recommendations to ensure reliable system operation. It also contains troubleshooting information. Table 1-1. Manual Conventions Element Convention Example Abbreviations Lowercase in. (inch) cm (centimeter) Acronyms Uppercase LED (Light-Emitting Diode) Model names First or formal reference: initial caps 3M Model 2101 Sensor Bar Subsequent use or informal reference: Initial caps for Model, lowercase for remainder Model 2101 sensor bar or sensor bar Copyright © 2001, 3M IPC. All rights reserved. 2 2 3M™ Vehicle System Series 2000VS Installation Safety Information We provide important safety information and warnings to assist you in understanding and avoiding potential harm to yourself, and possible damage to equipment, during the installation of the 3M Vehicle System Series 2000VS. Although we have included potential hazards you may encounter during the installation of this product, we cannot predict all of the possible hazards and this list should not be a substitute for your judgment and experience. Please read and observe all safety information and instructions in this manual before installing the system equipment. Also, save this installation manual and keep it for future reference. If you are unsure about any part of this installation or the potential hazards discussed, please contact your supervisor immediately. 2.1 Intended Use Statement The system is intended to make operators aware of the snowplow’s location with respect to the magnetic lane marking tape during low visibility caused by inclement weather conditions. 2.2 The safety message box contains a safety alert symbol ( ), the signal word WARNING, and a safety message. WARNING The safety message is in this box. WARNING means you and/or someone else MAY be KILLED or SERIOUSLY HURT if you do not follow these instructions. The following safety messages appear in this manual: WARNING Vehicle batteries contain sulfuric acid and may contain explosive gases. Keep sparks, flames, and cigarettes away. Wear eye protection. Disconnect the negative cable first to prevent shorting the positive terminal to the chassis when removing the positive cable. Battery acid may cause skin irritation and eye injury. Explosive gases may cause severe injury or death. Technical Support If you have questions about the system, its use, or operation, please contact your dealer or call 3M Intelligent Transportation Systems Technical Service department at 1-800-258-4610. 2.3 Safety messages are designed to alert you to potential hazards that can cause personal injury to you or others. Safety Messages We include safety messages and safety labels in this manual to help you protect your safety and the safety of others. This section contains important information to help you recognize and understand these safety messages. Please read all messages before proceeding with the installation. Copyright © 2001, 3M IPC. All rights reserved. WARNING A completed installation that is not tested may result in improper system operation, which may result in accidents and injuries. To avoid this problem, test the system to verify proper operation. Improper system operation may result in unsafe driver action. WARNING Do not use the operator display as the sole guidance of the vehicle. Make sure you can see the road in front of the vehicle. Always drive in a safe manner for existing road conditions. Overdriving road conditions may result in accidents and injuries. 3M™ Vehicle System Series 2000VS Installation 2.4 Safety Considerations Please consider the following safety issues before beginning the installation. Although we have compiled this list of common safety considerations, it should not be considered as complete. It is not intended to take the place of your good judgment, training, and experience. 2.4.1 Personal Safety Equipment and Clothing Personal safety equipment and clothing including high visibility vests, hard hats, gloves, electrical shock or electrocution protection clothing and equipment, safety shoes, safety glasses, face shields, goggles, and hearing protection devices are just some of the items available to you. Choose the right equipment for the job. If you are unsure of which safety equipment is recommended or appropriate for the job, ask your supervisor or foreman. 2.4.2 Electric Shock As a trained installer of electrical equipment you are aware of the dangers associated with installation of electrical devices. Always be sure that power to the equipment, and all associated equipment, is turned off and the vehicle battery is disconnected. Use equipment, techniques, and procedures that you learned during your training or apprenticeship or other electrical industry recognized safety procedures. 3 2.4.3 Explosion Common truck-type batteries produce an explosive gas under some conditions. This gas may easily be ignited by a spark or flame as you work on the vehicle. To reduce the risk of explosion, disconnect the battery, work in a well-ventilated area, avoid the use of devices that create sparks or use open flames, and use the appropriate personal safety equipment and clothing. If you are unsure of which techniques, procedures, and protective equipment are recommended or appropriate for the job, ask your supervisor or foreman. 2.4.4 Chemical Burns Common truck-type batteries contain strong acid that can cause personal injury if you come in contact with the acid. To reduce exposure to the risk of chemical burns, wear appropriate protective clothing and handle the battery with care. If you are unsure of which techniques, procedures, and protective equipment are recommended or appropriate for the job, ask your supervisor or foreman. 2.5 Disposal of Product Please dispose of the product and installation materials in accordance with all local, state, and federal laws and regulations. If you are unsure of which techniques, procedures, and protective equipment are recommended or appropriate for the job, ask your supervisor or foreman. Copyright © 2001, 3M IPC. All rights reserved. 4 3 3M™ Vehicle System Series 2000VS Installation Description This section provides a general description of the 3M™ Lane Awareness System and a detailed description of the Vehicle System Series 2000VS components. 3.1 3.2 The vehicle system consists of the following matched components: • Model 2100 Sensing Electronics Kit: Model 2101 Sensor Bar Model 2102 Sensor Bar Cable Model 2103 Standard Sensor Mount Kit • Model 2130 Junction Box Kit: Model 2131 Junction Box Model 2132 Speed Input Cable • Model 2500 Display Kit: Model 2501 Display Model 2502 Display Cable Model 2503 Display Floor Mounting Kit Lane Awareness System The 3M lane awareness system consists of a continuous magnetic pavement marking tape along the center or edge of the road, a sensor bar mounted on the front bumper of a snowplow truck, an onboard operator display, and an operator-alert system. The sensor bar detects the presence of the tape and transmits information to the operator display, indicating where the sensor bar is located in relation to the magnetic tape. If the sensor bar is outside the accepted operating range, the operator is alerted by vibration in the seat. If the optional peripheral vision lights are installed, the operator is also alerted by flashing lights positioned in his or her peripheral vision. The combination of these components allows snowplow operators to maintain even speeds with fewer slowdowns, avoid roadside obstacles, and clear roads quickly even during severe winter conditions. Vehicle System Series 2000VS Also Available: Model 2504 Display Dash Mounting Kit Model 2505 Display Windshield Mounting Kit • Model 2500M Display Kit: Model 2501M Display Metric Model 2502 Display Cable Model 2503 Display Floor Mounting Kit Also Available: Model 2504 Display Dash Mounting Kit Model 2505 Display Windshield Mounting Kit Copyright © 2001, 3M IPC. All rights reserved. • Model 2510 Vibration Seat Kit: Includes vibration seat motors and cable connector. • Model 2520 Peripheral Vision Light Kit: (set of two; these lights are optional) Model 2521 Peripheral Vision Light (single) Model 2522 Replacement LED 3M™ Vehicle System Series 2000VS Installation 5 Sensor Bar Junction Box One or two sensor bars are mounted to the front bumper of a snowplow truck using either standard mounting brackets or modified mounting brackets. The standard mounting brackets are intended for use on a heavy-duty dump truck equipped with a standard C-channel front bumper. The sensor bars and mounting brackets are heavy-duty, weather and corrosion-resistant components designed for years of trouble-free operation. The junction box is mounted inside the cab of the truck. It provides a convenient connection and distribution point for the sensing and power cables. The junction box is hinged for easy access to the cable connectors. The sensor bar can detect magnetic pavement marking tape up to 3 feet away (left 3 feet and right 3 feet from the center of the tape). The sensor bar converts detected magnetic energy into electrical signals that are transmitted through the sensor bar cable to the operator display. Operator Display The operator display indicates the position of the sensor bar in relation to the tape. The display is a compact, self-contained unit that is mounted inside the cab of the truck. It can be mounted either on a pedestal attached to the floor, on a bracket attached to the dash, or on a bracket attached to the windshield. The display has a Power button that latches in the ON position, a switch to select the left or right sensor bar, and control knobs to adjust the brightness of the display and the intensity of the vibration seat and peripheral vision lights. It also has control knobs to set the operating range (lateral distance that the sensor bar can drift away from the magnetic tape before the operator is alerted). Refer to Section 5, Checkout, for information on how to set up and test the operator display. Vibration Seat and PVLs Prior to driving the vehicle over the tape, the operator selects the left or right sensor bar and sets the desired operating range using the controls located on the front of the operator display. If the operator drives so that the sensor bar drifts outside the specified operating range, the operator is alerted by one side of the seat vibrating. If the optional peripheral vision lights are installed, the corresponding light flashes while the seat vibrates. To get back on track, the operator drives away from the seat vibration and flashing light. In other words, when the left side of the seat vibrates, the operator steers right; when the right side of the seat vibrates the operator steers left. When the sensor bar is back within the specified operating range, the seat stops vibrating and the light stops flashing. Figure 3-1 shows a typical installation of 3M™ Vehicle System Series 2000VS components. The right sensor bar and mounting bracket are hidden from view by the snowplow. Copyright © 2001, 3M IPC. All rights reserved. 6 3M™ Vehicle System Series 2000VS Installation Figure 3-1. Typical System Installation for Snow Removal Copyright © 2001, 3M IPC. All rights reserved. 3M™ Vehicle System Series 2000VS Installation 4 7 Installation This section describes the installation of 3M™ Vehicle System Series 2000VS components. 4.1 Pre-Installation Vehicle Inspection A 3M representative will inspect the vehicle to determine the installation requirements for that particular vehicle. The 3M representative will then make recommendations to facilitate an efficient and practical installation of vehicle system series 2000VS components. Copyright © 2001, 3M IPC. All rights reserved. 8 4.2 3M™ Vehicle System Series 2000VS Installation Installation Procedures Please read and fully understand the following precautionary paragraphs before starting the installation. WARNING Vehicle batteries contain sulfuric acid and may contain explosive gases. Keep sparks, flames, and cigarettes away. Wear eye protection. Disconnect the negative cable first to prevent shorting the positive terminal to the chassis when removing the positive cable. Battery acid may cause skin irritation and eye injury. Explosive gases may cause severe injury or death. • Before cutting or drilling any openings in the vehicle, draw a diagram showing placement, measurements, and dimensions. Use the diagram to avoid drilling or cutting holes in undesirable locations. • Always follow the vehicle manufacturer’s recommendations concerning modification, alteration, and installation or connection of accessories or equipment to the vehicle. It is the installer’s responsibility to ensure that attachment points are free and clear from any truck wiring, hoses, or other items that may in anyway affect other critical systems of the vehicle operation. 1. Disconnect the battery cables from the battery before beginning the installation. Disconnect the negative battery cable first, then the positive battery cable. • Cables that are routed under floor mats should be run between the pad and the mat to minimize abrasion or heat damage. 3. Install the vehicle system series 2000VS components in the following order: • Protect cables with armor or sheathing when they are routed around sharp corners and edges. Avoid routing cables through potential pinch points. Clamp or tie all cables in place. Route and secure cables well away from moving parts. • • Follow the installation instructions to avoid possible occurrence of false sensing or diminished sensing range. Perform a pre-installation vehicle inspection to make sure that the sensor mounting bracket can be mounted so the sensor bar is within specifications (see Subsection 4.2.5, Sensor Mounting Bracket Installation Guidelines). Also determine the best mounting locations for the operator display, junction box, and peripheral vision lights to avoid problems with existing operator equipment and controls. Copyright © 2001, 3M IPC. All rights reserved. 2. Remove interior panels, as necessary, to provide access for cable routing. Replace the panels when you are done routing cables. • Operator Display Mount • Peripheral Vision Lights (Optional) • Junction Box • Sensor Mounting Bracket • Sensor Bar • Operator Display Note These installation instructions assume that the vibration seat is already installed in the vehicle and only needs to be connected to the junction box. 3M™ Vehicle System Series 2000VS Installation 4.2.1 9 Operator Display Mount The 3M™ Model 2501 or 2501M Display can be mounted on one of three mounting brackets. Figure 4-1 shows an operator display and the available mounting brackets. • Floor-mounted pedestal with adjustable mounting plates. See Subsection 4.2.1.1, Display Floor Mounting Bracket Installation. • Dash-mounted bracket with flexible tubing. See Subsection 4.2.1.2, Display Dash Mounting Bracket Installation. • Windshield-mounted bracket with an adjustable base. See Subsection 4.2.1.3, Display Windshield Mounting Bracket Installation. Figure 4-1. Operator Display and Mounting Brackets Copyright © 2001, 3M IPC. All rights reserved. 10 3M™ Vehicle System Series 2000VS Installation 4.2.1.1 Display Floor Mounting Bracket Installation Figure 4-2 shows a 3M™ Model 2503 Display Floor Mounting Kit. This display mounting bracket is included with the Model 2500 or 2500M display kit. 1. Remove the floor mounting bracket and small bag of accessories from the packaging. An Allen wrench is provided with the mounting bracket. 2. Use the Allen wrench to loosen the four setscrews. Then separate the upper and lower assemblies from the short rod that holds them together. 3. Determine the best location to install the floor base plate. 4. Attach the base plate to the floor of the truck cab using the self-drilling/tapping hex-head screws provided. Note It may be necessary to remove the lower assembly from the floor base plate before attaching the plate to the floor of the truck. Do this by loosening the locking nut and unscrewing the assembly from the base plate. Be sure to tighten the locking nut when you reattach the lower assembly to the floor base plate. 5. Insert the extension rod into the lower assembly and tighten the two setscrews. 6. Place the upper assembly onto the extension rod. Rotate the display mounting plate to the desired position and tighten the two setscrews. Note The operator display and cables will be installed later. See Subsection 4.2.8, Operator Display Installation. Copyright © 2001, 3M IPC. All rights reserved. Figure 4-2. Display Floor Mounting Bracket 3M™ Vehicle System Series 2000VS Installation 4.2.1.2 11 Display Dash Mounting Bracket Installation Figure 4-3 shows a 3M™ Model 2504 Display Dash Mounting Kit with its base plate configured to mount to the front of the dash. Figure 4-4 shows the base plate configured to mount to the top of the dash. 1. Remove the dash mounting bracket and small bag of accessories from the packaging. Two Allen wrenches are provided with the mounting bracket. 2. Use the larger Allen wrench to loosen the setscrew and remove the flexible tubing from the dash base plate. 3. Determine the best location to install the dash base plate. Note Choose a location that will not adversely obstruct the driver’s view when the operator display and its associated cables are installed. 4. Attach the base plate to the dash of the truck cab using the self-drilling/tapping hex-head screws provided. Figure 4-3. Display Dash Mounting Bracket (Mount to Front of Dash) 5. Insert the rod end of the flexible tubing into the desired hole in the base plate. Rotate the display mounting plate to the desired position and tighten the setscrew. 6. Press the three large plastic caps into the holes in the base plate as shown. 7. Press the small plastic cap into the remaining hole in the base plate. Note The operator display and cables will be installed later. See Subsection 4.2.8, Operator Display Installation. Figure 4-4. Display Dash Mounting Bracket (Mount to Top of Dash) Copyright © 2001, 3M IPC. All rights reserved. 12 3M™ Vehicle System Series 2000VS Installation 4.2.1.3 Display Windshield Mounting Bracket Installation Figure 4-5 shows a 3M™ Model 2505 Display Windshield Mounting Kit. The back of the base plate has high-performance adhesive tape that sticks to the inside of the windshield. The windshield surface preparation and cleanliness is paramount to achieve the proper level of bonding. 1. Remove the windshield-mounting bracket from the packaging. 2. Unscrew the locking nut and separate the display mounting plate from the windshield base plate. 3. Determine the best location to install the windshield base plate. If possible, choose a location where the windshield is relatively flat. Try to avoid areas of the windshield that have a severe curvature. Note Choose a location that will not adversely obstruct the driver’s view when the operator display and its associated cables are installed. 4. Clean the inside of the windshield with a good quality window cleaner. Then, using the alcohol pad provided, wipe the area of the windshield where the base plate will be installed. 5. As soon as the windshield dries, peel the liner from the tape on the back of the base plate and press the base plate to the windshield at the desired location. Apply pressure on the windshield base plate for at least two minutes (longer is better) to ensure maximum bonding. Note Do not attempt to install the operator display at this time because the high-performance adhesive requires a curing time of at least 30 minutes. See Subsection 4.2.8, Operator Display Installation. Copyright © 2001, 3M IPC. All rights reserved. Figure 4-5. Display Windshield Mounting Bracket 3M™ Vehicle System Series 2000VS Installation 4.2.2 Peripheral Vision Lights Installation Figure 4-6 shows an optional 3M™ Model 2520 Peripheral Vision Light Kit. The back of each mounting pad has high-performance adhesive tape that sticks to the inside of the windshield. The windshield surface preparation and cleanliness is paramount to achieve the proper level of bonding. Attach Windshield Mounting Pads 1. Remove the peripheral vision lights from the packaging. 13 Attach Lights to Mounting Pads 1. After the adhesive on the mounting pads has cured for at least 30 minutes, attach the peripheral vision lights to the mounting pads by pressing the light assemblies firmly against the mounting pads. 2. Route the cables to the junction box and connect them to the corresponding connectors as shown in Figure 4-8, Junction Box Cable Connections. 3. Tie the cables so they do not obstruct the driver’s view of the windshield and they are away from any moving parts. 2. Determine the best locations to install the two windshield mounting pads. If possible, choose locations where the windshield is relatively flat. Try to avoid areas of the windshield that have a severe curvature. Note Choose locations that are within the drivers peripheral vision (one on the left and one on the right), but will not adversely obstruct the driver’s view when the LED assemblies and cables are installed. 3. Clean the inside of the windshield with a good quality window cleaner. Then, using the alcohol pads provided, wipe the two areas of the windshield where the mounting pads will be installed. 4. As soon as the windshield dries, peel the liner from the tape on the back of the mounting pads and press the mounting pads to the windshield at the desired locations. Apply pressure on the mounting pads for at least two minutes (longer is better) to ensure maximum bonding. Note Do not attempt to attach the lights at this time because the high-performance adhesive requires a curing time of at least 30 minutes. Figure 4-6. Peripheral Vision Lights Copyright © 2001, 3M IPC. All rights reserved. 14 4.2.3 3M™ Vehicle System Series 2000VS Installation Junction Box Installation Figure 4-7 shows a 3M™ Model 2131 Junction Box and the cables that are permanently attached to it. 1. Remove the junction box and cables from the packaging. 2. Determine the best location to install the junction box. Note Choose a location on the back wall of the truck cab with sufficient room below the junction box to route the power and signal cables. Also provide room to remove the cover to gain access to the inside of the box. 3. Attach the junction box to the back wall of the truck cab using the self-drilling/tapping hex-head screws provided. Orient the junction box so that the hinge is on the top and the cables extend out of the bottom. 4. Route the truck power cable to the truck’s power distribution point, but do not connect the wires yet. 5. Route the display power cable to the display mounting bracket. It will be connected later. 6. Route the vibration seat cable to the driver’s seat, locate the connector attached to the seat, and plug the two connectors together. Tie the cables to the seat to provide connector strain relief. 7. Tie all cables away from any moving parts. Figure 4-7. Junction Box Installation Copyright © 2001, 3M IPC. All rights reserved. 3M™ Vehicle System Series 2000VS Installation 4.2.4 15 Junction Box Cable Connections Figure 4-8 shows the power and signal cables connected to the junction box. Self-resetting fuses are located inside the junction box: one fuse for the sensor bars and display, and another fuse for the vibration seat motors. Figure 4-8. Junction Box Cable Connections Copyright © 2001, 3M IPC. All rights reserved. 16 4.2.5 3M™ Vehicle System Series 2000VS Installation Sensor Mounting Bracket Installation Guidelines The 3M™ Model 2103 Standard Sensor Mount Kit should be used for an installation where the truck is equipped with a standard C-channel bumper. For other bumper configurations, a non-standard sensor mounting bracket is required. In this case, a 3M representative will determine the needs and make recommendations based on the pre-installation vehicle inspection. The sensor bar’s optimal vertical distance above the tape is 12 inches. Its horizontal distance from the tire should be at least 12 inches to prevent magnetic noise caused by tire rotation from affecting performance. Figure 4-9 shows positional requirements for the sensor bar. Note If these installation guidelines are not followed, system performance may be degraded (for example, increased frequency of false alerts, inaccuracies or diminished sensing range). Figure 4-9. Sensor Bar Positional Requirements Copyright © 2001, 3M IPC. All rights reserved. 3M™ Vehicle System Series 2000VS Installation 4.2.6 Standard Sensor Mounting Bracket Installation 1. Remove the sensor mounting bracket and clamping hardware from the packaging. A T-handle 5 mm hex wrench is provided. 2. Check all fasteners on the pre-assembled sensor mounting bracket to ensure they are secure. 3. Thread one M10 hex-nut with lockwasher onto each of the four threaded rods. Then thread the rods into the lower clamping blocks securely and tighten the nuts. 4. Place this assembly under the bumper with the rods extending up on each side of the bumper as shown in Figure 4-10. 5. Place the upper clamping blocks over the threaded rods and secure them with two M10 hex-nuts and one lockwasher as shown. 17 7. Attach the sensor bar clamping blocks to the lower horizontal arm of the mounting bracket. a. Apply Loctite 262 to the threads of the four M8 x 30 bolts. b. Insert two bolts into each clamping block and thread a T-slot nut onto each bolt. Do not tighten the nuts. c. Orient the clamping blocks so the round bosses will be facing each other when they are installed. d. Slide the slotted clamping block into the sensor mounting bracket. Make sure the open part of the slot is pointing down. e. Slide the other clamping block into the sensor mounting bracket. Do not tighten the bolts. 8. Replace the end cap and secure it with the screws and washers previously removed. 6. Remove the end cap from the lower horizontal arm of the sensor mounting bracket as shown. Figure 4-10. Standard Sensor Mounting Bracket Copyright © 2001, 3M IPC. All rights reserved. 18 4.2.7 3M™ Vehicle System Series 2000VS Installation Sensor Bar and Cable Installation 1. Remove the sensor bar and cable from the packaging. 2. Remove the protective cap, if present, from one end of the cable and connect the cable to the sensor bar. Do not over tighten the connector; finger tight will provide the proper seal. Note The sensor bar has the letter B stamped on the bottom near the cable connector. 3. Orient the sensor bar so the cable end will be connected to the slotted clamping block and the letter B will be facing down as shown in Figure 4-11. 4. Hold the sensor bar in the desired position and slide the two clamping blocks firmly against the bar. Make sure the round bosses on the clamping blocks interlock with the sensor bar. Then tighten the bolts securing the clamping blocks using a 5 mm hex wrench. 5. Ensure that the sensor bar is securely clamped to the mounting bracket. Grab onto the sensor bar and forcibly try to move it. If you feel any motion, tighten the clamping blocks and retest until the sensor bar is secure. 6. Tie the cable to the sensor mounting bracket. Then route the cable to the junction box. Avoid sharp bends and do not kink the cable. Pass the cable through an existing engine compartment firewall access port if available. Otherwise, punch a new hole in the firewall and pass the cable through the hole. Note Tie the cable to avoid pinch points and moving parts especially in the engine compartment. Try to minimize the possibility of signal corruption by routing the cable away from high electrical noise sources, such as the alternator or engine ignition system. 7. Connect the sensor bar cable to the corresponding left or right connector as shown in Figure 4-8, Junction Box Cable Connections. Figure 4-11. Sensor Bar and Cable Installation Copyright © 2001, 3M IPC. All rights reserved. 3M™ Vehicle System Series 2000VS Installation 4.2.8 Operator Display Installation Figure 4-12 shows a 3M™ Model 2501 Display with its signal and power cables and the various mounting brackets. 1. Align the holes in the back of the operator display with the holes in the mounting plate and secure the display to the mounting plate with the screws and washers provided. If the windshield mounting bracket is used, secure the display and mounting plate to the windshield base plate using the locking nut. 19 2. Connect the display power cable to the operator display. This cable is already connected to the junction box. 3. Connect one end of the display signal cable to the operator display. 4. Route the display signal cable to the junction box and connect it as shown in Figure 4-8, Junction Box Cable Connections. 5. Tie the cables so they do not obstruct the driver’s view of the windshield and they are away from any moving parts. Figure 4-12. Operator Display Installation Copyright © 2001, 3M IPC. All rights reserved. 20 4.2.9 3M™ Vehicle System Series 2000VS Installation Speed Input Cable Installation 1. Remove the junction box cover to gain access to the terminal block. 7. Route the speed input cable to the vehicle’s speed input source and connect the wires to the proper connection points. Note 2. Loosen the cable strain relief. See Figure 4-13. 3. Feed the speed input cable through the strain relief so that the wires easily reach the terminal block. Tighten the cable strain relief. 4. Strip 1/4 inch of insulation from the red and black wires. 5. Locate the correct terminal for the type of speed pickup sensor and connect the red wire to (+) and the black wire to (–). Table 4-1 lists the terminal to use for the corresponding speed input type. 6. Replace the junction box cover and secure it with the screws previously removed. Tie the cable to avoid pinch points and moving parts especially in the engine compartment. Try to minimize the possibility of signal corruption by routing the cable away from high electrical noise sources, such as the alternator or engine ignition system. Table 4-1. Terminals for Speed Input Connection Terminal J10 J1708 INPUT J11 VEL1 INPUT J12 VEL2 INPUT J13 Not used J14 Not used J15 Not used Figure 4-13. Speed Input Cable Installation Copyright © 2001, 3M IPC. All rights reserved. Speed Input Type 3M™ Vehicle System Series 2000VS Installation 21 4.2.10 Cable Conduit Installation Figure 4-14 shows the junction box with the cables bundled in conduit. The conduit will provide added protection for the cables. 1. Cut the cable conduit to the desired length. 2. Insert the cables into the conduit. 3. Use cable ties to secure the cable bundle. 4. Cut off excess cable tie lengths. Figure 4-14. Junction Box With Cables Bundled in Conduit Copyright © 2001, 3M IPC. All rights reserved. 22 3M™ Vehicle System Series 2000VS Installation 4.2.11 Truck Power Connections WARNING Vehicle batteries contain sulfuric acid and may contain explosive gases. Keep sparks, flames, and cigarettes away. Wear eye protection. Disconnect the negative cable first to prevent shorting the positive terminal to the chassis when removing the positive cable. Battery acid may cause skin irritation and eye injury. Explosive gases may cause severe injury or death. 1. At the truck’s power distribution point, locate a source of +12 VDC power that is active only when the vehicle key is in the RUN position. 2. Test the power source to make sure it provides adequate current carrying capacity (10 amps minimum). 3. Connect the red power wire to the +12 VDC power source. 4. Connect the black wire to a (–) terminal that is connected directly to the vehicle chassis ground. 5. Connect the vehicle’s battery cables to the battery. Connect the positive battery cable first, then connect the negative battery cable. Copyright © 2001, 3M IPC. All rights reserved. 3M™ Vehicle System Series 2000VS Installation 5 Checkout This section describes how to set up and test the operator interface components and sensor bars. 5.1 Installation Checkout 1. Turn the vehicle key to the RUN position. 2. Push the Power button to turn on the operator display. See Figure 5-1. The display turns on and the system initiates a brief test of the selected sensor bar to verify proper operation. Refer to the operator display icons shown in Figure 5-2. 3. If two sensor bars are installed, select the other sensor bar using the Sensor Bar Selector switch. The system initiates a brief test of the sensor bar to verify proper operation. 4. Push the Vibration Seat button once to start a 30-second test of the seat. Push the button again to stop the test. Both sides of the seat will vibrate for 30 seconds or until stopped. While the seat is vibrating, set the desired intensity of the vibration by turning the Vibration Seat button (CW increases intensity, CCW decreases intensity). 23 5. Push the Peripheral Vision Lights button once to start a 30-second test of the lights. Push the button again to stop the test. Both lights will flash for 30 seconds or until stopped. While the lights are flashing, set the desired intensity of the lights by turning the Peripheral Vision Lights button (CW increases intensity, CCW decreases intensity). 6. Set the desired brightness of the operator display by turning the Brightness button (CW increases brightness, CCW decreases brightness). 7. Push the Brightness button once to start a 5-minute test of the system. Push the button again to stop the test. In test mode, the display simulates drifting in and out of the operating range as shown in Figure 5-3. As the arrow representing the sensor bar drifts outside the operating range to the left, the left side of the seat vibrates and the left light flashes. As the arrow drifts outside the operating range to the right, the right side of the seat vibrates and the right light flashes. 8. Set the desired operating range by turning the Operating Range Control knobs. The range indicators move as you turn the knobs. Figure 5-1. Operator Display Copyright © 2001, 3M IPC. All rights reserved. 24 3M™ Vehicle System Series 2000VS Installation Figure 5-2. Operator Display Icons Figure 5-3. Test Mode Active for Left Sensor Bar Copyright © 2001, 3M IPC. All rights reserved. 3M™ Vehicle System Series 2000VS Installation 5.2 Performance Test 25 Set Up Operator Display 1. Turn the vehicle key to the RUN position. WARNING A completed installation that is not tested may result in improper system operation, which may result in accidents and injuries. To avoid this problem, test the system to verify proper operation. Improper system operation may result in unsafe driver action. These installation instructions are the result of tests performed in our laboratory and we believe these tests to be accurate and complete. However, each installation involves variables that cannot be controlled or predicted. These variables may affect the operational characteristics of the system. To ensure proper system operation, 3M strongly recommends that the installer functionally test the system by setting up the operator display and driving the vehicle with the selected sensor bar positioned over magnetic pavement marking tape. WARNING Do not use the operator display as the sole guidance of the vehicle. Make sure you can see the road in front of the vehicle. Always drive in a safe manner for existing road conditions. Overdriving road conditions may result in accidents and injuries. 2. Push the Power button on the operator display. The display turns on and the system initiates a test of the selected sensor bar to verify proper operation. Refer to the operator display icons shown in Figure 5-2. 3. Select the desired sensor bar using the Sensor Bar Selector switch. 4. Set the desired operating range by turning the left and right Operating Range Control knobs. The range indicators move as you turn the knobs. 5. Push the Power button to turn off the operator display until you are ready to drive the vehicle with the sensor bar positioned over tape. Drive Vehicle With Sensor Bar Over Tape 1. With the vehicle running, push the Power button on the operator display to turn it on. The display shows the operating range you previously set. 2. Drive the vehicle with the selected sensor bar positioned over the tape. Try to stay within the range you set. 3. If you drive so that the sensor bar drifts outside the selected operating range, the left or right side of the seat vibrates: When the left side vibrates, steer right. When the right side vibrates, steer left. If the optional peripheral vision lights are installed, the corresponding light flashes while the seat vibrates. 4. Verify that the system is performing properly. If the system is not performing properly, begin appropriate systematic troubleshooting procedures to correct the malfunction. Refer to the troubleshooting procedures in Section 6, Maintenance and Troubleshooting. Copyright © 2001, 3M IPC. All rights reserved. 26 6 3M™ Vehicle System Series 2000VS Installation Maintenance and Troubleshooting 6.1.1 Sensor Bar Storage The 3M™ Vehicle System Series 2000VS components are designed for reliable operation. Inspect the components at regular intervals to ensure proper system operation. 1. Loosen the bolts securing the clamping blocks using a 5 mm hex wrench. If the bolts are dirty and difficult to loosen, saturate them with penetrating oil. 3M recommends the following: 2. Hold onto the sensor bar and slide the clamping blocks away from the ends of the sensor bar. See Figures 4-10 and 4-11. • • 6.1 The sensor bar(s) and mounting bracket(s) should be inspected before each use of the system to make sure that the mounting hardware has not vibrated loose. If anything is loose, tighten it. Turn on the operator display and test the system components as described in Section 5, Checkout. Do this before each use of the system. If any problems occur, refer to Subsection 6.2, Troubleshooting. Off-Season Component Storage When the snow season is over, remove the sensor bar(s) and mounting hardware. Also remove the operator display. Store the components in an appropriate place where they will be safe until the next snow season. 3. Disconnect the cable from the sensor bar and thread a protective cap onto the end of the cable. Then cut the cable ties securing the sensor bar cable to the sensor mounting bracket. 4. Remove the upper clamping blocks that secure the sensor mounting bracket to the bumper and remove the bracket. See Figure 4-10. 5. Coil the sensor bar cable inside of the engine compartment and tie the cable away from moving parts. 6. Put the sensor bar(s) and mounting hardware in storage. Note If available, put each sensor bar in its shipping carton. Otherwise, wrap each bar with packing material to protect it. Store the sensor bar(s) away from strong magnetic fields. Note Do not remove the junction box. Leave the cables connected to the junction box. 6.1.2 Operator Display Storage 1. Disconnect the power and signal cables from the side of the display. Tie the cables away from moving parts. 2. Disconnect the display from its mounting plate. Keep the mounting hardware with the display. 3. Wrap the display with packing material to protect the screen and control knobs. Then put the display in storage. Copyright © 2001, 3M IPC. All rights reserved. 3M™ Vehicle System Series 2000VS Installation 6.2 27 Troubleshooting Refer to the following tables for troubleshooting symptoms and solutions: • Table 6-1, Operator Display Troubleshooting • Table 6-2, Vibration Seat Troubleshooting • Table 6-3, Peripheral Vision Lights Troubleshooting Table 6-1. Operator Display Troubleshooting Symptom/Indication Operator display is blank. Possible Cause Solution Power switch is OFF. Push Power button on display so it latches in ON position. Display power cable not connected properly. Check display power cable connection. Make sure cable is plugged into side of operator display. Unplug cable and plug it in again. Brightness control set to minimum. Turn Brightness knob clockwise to brighter setting. No power from vehicle. Turn vehicle key to RUN position. Check green light (LED) in junction box. If green LED is ON, power to junction box is OK. See Figure 4-13 for location of green LED. Check truck power cable connection. Make sure red power wire from junction box is connected to +12 VDC power source that is active only when vehicle key is in RUN position. Make sure black wire is connected to (–) terminal that is connected directly to vehicle chassis ground. No Communication icon displayed. Sensor Bar Selector switch set to wrong sensor bar. Toggle Sensor Bar Selector switch to correct sensor bar. Sensor bar cables connected incorrectly. Make sure sensor bar cables are connected to correct connectors at junction box as shown in Figure 4-8, Junction Box Cable Connections. Loose display signal cable connection. Tighten display signal cable connectors at display and at junction box. Display failure. Replace operator display and retest. Return unit to 3M for service. Sensor bar cable failure. Replace sensor bar cable and retest. Sensor bar failure. Replace sensor bar and retest. Return unit to 3M for service. Junction box failure. Replace junction box and retest. Return unit to 3M for service. Copyright © 2001, 3M IPC. All rights reserved. 28 3M™ Vehicle System Series 2000VS Installation Table 6-1. Operator Display Troubleshooting (continued) Symptom/Indication Possible Cause Solution Stray magnetic field caused system malfunction or system was started over tape. Restart system by turning display OFF, then ON again. Sensor bar failure. Replace sensor bar and retest. Return unit to 3M for service. Speed input signal from vehicle lost. Check speed input cable connection in junction box. See Subsection 4.2.9, Speed Input Cable Installation. Speed input cable failure. Replace speed input cable and retest. Speed input signal from vehicle is faulty. Contact 3M representative. No Tape icon displayed when sensor bar is over tape. Sensor Bar Selector switch set to wrong sensor bar. Toggle Sensor Bar Selector switch to correct sensor bar. Vehicle speed too slow. Repeat observation at speed greater than 5 mph. Tape signal present when sensor bar is NOT over tape. Magnetic noise caused by tire rotation. Reposition sensor bar. See Subsection 4.2.5, Sensor Mounting Bracket Installation Guidelines. Cables routed incorrectly. Reroute cables and retest. Display not tracking tape consistently. Magnetic noise caused by tire rotation. Reposition sensor bar. See Subsection 4.2.5, Sensor Mounting Bracket Installation Guidelines. Cables routed incorrectly. Reroute cables and retest. Speed input signal from vehicle is faulty. Contact 3M representative. Moisture condensed on display screen. Remove display and place it in dry area until clear. Then install display in truck and retest. Failed Test icon displayed when self-check runs. Speed Input Lost icon displayed. Display fogged up. Copyright © 2001, 3M IPC. All rights reserved. Stop vehicle, then restart system with vehicle in Park (Neutral). Turn truck ignition OFF, then ON again. Restart system. 3M™ Vehicle System Series 2000VS Installation 29 Table 6-2. Vibration Seat Troubleshooting Symptom/Indication Vibration seat test doesn’t work. Possible Cause Solution Vibration Seat control set to minimum. Turn Vibration Seat knob clockwise to higher setting. Display not working. Make sure display power cable is connected properly. No power to seat motors. Check vibration seat cable connection. Make sure cable connector is plugged into mating cable connector under driver’s seat. Connectors are keyed so they cannot be connected incorrectly. Vibration Seat button on display may have failed. Replace operator display and retest. Return unit to 3M for service. Seat vibrates less. Seat motor(s) failed. Contact 3M representative. Both sides of seat vibrate at the same time. Vibration seat is in test mode. Push Vibration Seat button to stop test mode. If vibration cannot be stopped, disconnect power to seat and contact 3M representative. Seat vibrates constantly. Operating range set incorrectly. Reset operating range using Operating Range control knobs on display. Display failure. Replace operator display and retest. Return unit to 3M for service. Table 6-3. Peripheral Vision Lights Troubleshooting Symptom/Indication Peripheral vision lights test doesn’t work. Possible Cause Solution PVL control set too low. Turn PVL knob clockwise to higher setting. Display not working. Make sure display power cable is connected properly. No power to PVLs. Check cable connection between PVLs and junction box. PVL button on display may have failed. Replace operator display and retest. Return unit to 3M for service. Left PVL flashes when right PVL should flash, and vice versa. PVL cables connected to wrong connectors at junction box. Make sure PVL cables are connected to correct connectors at junction box as shown in Figure 4-8, Junction Box Cable Connections. Both PVLs flash at the same time. Peripheral vision lights are in test mode. Push PVL button to stop test mode. If flashing cannot be stopped, disconnect power to lights and contact 3M representative. PVLs flash constantly. Operating range set incorrectly. Reset operating range using Operating Range control knobs on display. Display failure. Replace operator display and retest. Return unit to 3M for service. Improper cleaning of window. Clean window surface and reapply mounting pad. If necessary, order new mounting pad from 3M. See Subsection 4.2.2, Peripheral Vision Lights Installation. PVL fell off window. Copyright © 2001, 3M IPC. All rights reserved. 30 3M™ Vehicle System Series 2000VS Installation This page intentionally left blank. Copyright © 2001, 3M IPC. All rights reserved. Important Notice to the Purchaser THE FOLLOWING IS MADE IN LIEU OF ALL WARRANTIES OR CONDITIONS, EXPRESS OR IMPLIED, INCLUDING ANY IMPLIED WARRANTY OR CONDITION OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE: 3M shall replace or refund the purchase price of such quantity of product found to be defective in material or manufacture or not in conformance with 3M’s written specifications. Except as herein provided, 3M shall not be liable in contract or in tort for any injury, loss or damage, whether non-specified direct, indirect, incidental, special or consequential, arising out of the use, misuse or the inability to use the 3M™ Lane Awareness System for snow removal operations (or any of its component parts), regardless of the legal theory asserted. No statement or recommendations not contained herein shall have any force or effect unless in an agreement signed by officers of seller and manufacturer. THE REMEDIES SET FORTH HEREIN ARE EXCLUSIVE. Intelligent Transportation Systems 3M Safety and Security Systems Division 3M Canada Company 3M Center, Bldg. 225-4N-14 St. Paul, Minnesota 55144-1000 Customer Service 1-800-328-7098 Technical Service 1-800-258-4610 Worldwide Technical Service 1-651-575-5072 Website http://www.3M.com/its P.O. Box 5757 London, Ontario, Canada N6A 4T1 519-451-2500 1-800-3MHELPS Printed in U.S.A. Copyright © 2001, 3M IPC. All rights reserved. 75-0500-4532-9 Appendix E - Altra Technologies Side Radar Communication Specification. Message format from sensor to vehicle computer: Byte Description Number 0 Message number (23) 1 Sensor unit number (1 to 15) 2 Unused (0) 3 Ignore 4 Ignore 5 Ignore 6 Ignore 7 Ignore 8 Ignore 9 Ignore 10 Ignore 11 Ignore 12 Ignore 13 Ignore 14 Ignore 15 Ignore 16 Ignore 17 Priority Level (1= no alarm, 2=far alarm, 3=close alarm) 18 Unused (0) 19 Unused (0) 20 Unused (0) 21 Checksum of bytes 0 to 20 (exclusive OR) Table E1: Message format from sensor to vehicle computer. 1 Appendix F Routes and Randomization for Human Factors Experiments 1 and 2 using the HFRL Driving Simulator. Table F1 below indicates the routes for the first set of driving experiments undertaken in the HFRL driving simulator. Route Number Lane Width Segment #1 1 9 ft — Segment #2. [With distance of stationary vehicle from intersection & lane placement of stationary vehicle] — 2 10 ft — — 3 11 ft — 4 10 ft — 5 11 ft — 0.4 km (Out of lane) 1.6 km (Half in lane) — 6 9 ft — — Segment #3. [With distance of stationary vehicle from intersection & lane placement of stationary vehicle] — 0.4 km (Half in lane) — 0.8 km (Out of lane) — 0.4 km (Out of lane) Segment #4. [With distance of stationary vehicle from intersection & lane placement of stationary vehicle] 1.2 km (Out of lane) — 1.2 km (Half in lane) — 0.8 km (Half in lane) — Table F1: Routes used in Experiment 1 (with lane widths and location and placement of stationary vehicles). 1 Segment #5. [With distance of stationary vehicle from intersection & lane placement of stationary vehicle] 0.8 km (Half in lane) 1.6 km (Out of lane) — — 1.6 km (Out of lane) 1.2 km (Half in lane) Table F2 below indicates the randomization for the first set of driving experiments undertaken in the HFRL driving simulator. Subjects S1 S2 S3 S4 S5 S6 Route # Lane departure condition Collision warning condition Route # Lane departure condition Collision warning condition Route # Lane departure condition Collision warning condition Route # Lane departure condition Collision warning condition Route # Lane departure condition Collision warning condition Route # Lane departure condition Collision warning condition Trial 1 1 Nowarning Yellow Red Trial 2 3 Red Area & Red Only Trial 3 2 Double Line Trial 4 5 Red Line Trial 5 6 Area Nowarning Red Only Yellow Red 5 Red Line 2 Red Area 1 Double Line 3 Area 4 Nowarning Red Only Nowarning Red Only Yellow Red 3 Area 2 Double Line 1 Nowarning 5 Red Area Nowarning Red Only Yellow Red 4 Double Line 1 Area 3 Red Line 2 Red Area 5 Nowarning Yellow Red & Nowarning Red Only Nowarning Yellow & Red 6 Red Line 2 Area 1 Red Area 3 Double Line 4 Nowarning Nowarning Red Only Nowarning Yellow Red 4 Nowarning 3 Red Line 2 Red Area 1 Area 5 Double Line Red Only Nowarning Yellow Red & Red Only Yellow Red 2 & Nowarning & & Nowarning 6 Red Line Yellow Red & & Red Only & S7 S8 S9 S10 S11 S12 S13 Route # Lane departure condition Collision warning condition Route # Lane departure condition Collision warning condition Route # Lane departure condition Collision warning condition Route # Lane departure condition Collision warning condition Route # Lane departure condition Collision warning condition Route # Lane departure condition Collision warning condition Route # Lane departure 6 Red Area 4 Nowarning Nowarning Yellow Red 5 Area 6 Nowarning Yellow Red & Red Only 1 Red Line 3 Double Line 2 Area & Red Only Yellow Red & Nowarning 2 Double Line Yellow Red 1 Red Line & Nowarning 3 Red Area Red Only 6 Double Line 4 Nowarning 3 Area 2 Red Area 1 Red Line Red Only Yellow Red & Nowarning Red Only Nowarning 5 Red Line 6 Area 4 Nowarning 1 Double Line 3 Red Area Nowarning Red Only Nowarning Yellow Red & Red Only 4 Nowarning 5 Red Line 6 Double Line 2 Area Red Only Nowarning Yellow Red & Red Only Yellow Red 2 Red Area 5 Red Line 6 Nowarning 4 Area 1 Double Line Nowarning Yellow Red 3 Area 4 Red Area 3 & Red Only 5 Red Line Yellow Red 6 Nowarning 1 Red Area & Nowarning 2 Double Line & S14 S15 condition Collision warning condition Route # Lane departure condition Collision warning condition Route # Lane departure condition Collision warning condition Yellow Red & Red Only 3 Double Line 2 Red Area Red Only Yellow Red 1 Red Line 3 Double Line Yellow Red & Nowarning Yellow Red & Nowarning Red Only 5 Area 6 Red Line 4 Nowarning & Nowarning Red Only Nowarning 4 Red Area 5 Nowarning 6 Area Red Only Nowarning Yellow Red Table F2: Combination of route number, lane departure condition, and collision warning condition experienced in each trial by each subject. 4 & Table F3 below indicates the randomization of the tests performed by the snowplow operators in the second driver simulator experiment. Subject S1 Practice S2 S3 Practice Practice S4 S5 Practice Practice S6 S7 Practice Practice S8 S9 Practice Practice S10 S11 Practice Practice S12 S13 Practice Practice S14 S15 Practice Practice S16 S17 Practice Practice S18 S19 Practice Practice S20 S21 Practice Practice S22 S23 Practice Practice S24 S111 S112 Practice Practice Practice No HUD NoHUD No HUD NoHUD No HUD NoHUD No HUD NoHUD No HUD NoHUD No HUD NoHUD No HUD NoHUD No HUD NoHUD No HUD NoHUD No HUD NoHUD No HUD NoHUD No HUD NoHUD NoHUD No HUD HUDOnly 22(V) 33(A) 44(H) 58(VAH) HUDOnly 25(VA) HUDOnly 23(A) 37(HV) 48(VAH) 56(AH) 38(VAH) 42(V) 54(H) HUDOnly 27(HV) HUDOnly 24(H) 36(AH) 32(V) 45(VA) 58(VAH) 38(VAH) 53(A) HUDOnly 26(AH) 38(VAH) 47(HV) HUDOnly 28(VAH) 34(H) 43(A) HUDOnly 28(VAH) 35(VA) HUDOnly 22(V) 33(A) HUDOnly 25(VA) HUDOnly 23(A) 55(VA) 52(V) 46(AH) 57(HV) 48(VAH) 54(H) 38(VAH) 46(AH) 34(H) 42(V) 57(HV) 58(VAH) HUDOnly 28(VAH) 37(HV) 45(VA) HUDOnly 24(H) 38(VAH) 43(A) 56(AH) 52(V) HUDOnly 26(AH) 35(VA) HUDOnly 28(VAH) 32(V) 47(HV) 44(H) 58(VAH) 53(A) HUDOnly 27(HV) HUDOnly 22(V) 36(AH) 34(H) 48(VAH) 55(VA) 43(A) 58(VAH) HUDOnly 25(VA) HUDOnly 24(H) 38(VAH) 47HV) 38(VAH) 42(V) 56(AH) 53(A) HUDOnly 28(VAH) 36(AH) HUDOnly 23(A) 32(V) 45(VA) 57 48(VAH) 54(H) HUDOnly 26(AH) 37(HV) HUDOnly 28(VAH) 33(A) 48(VAH) 55(VA) 44(H) 52(V) HUDOnly 27(HV) HUDOnly 23(A) HUDOnly 27(HV) 35(VA) 46(AH) 32(V) 44(H) 38(VAH) 46(AH) 58(VAH) 58(VAH) 55(VA) Table F3: Order in which conditions were presented to Snowplow Operators. 5 Table F4 below indicates the randomization of the tests performed by the state patrol officers in the second driver simulator experiment. Subject H1 Practice 1HUD On H2 H3 1HUDOnly 22(V) 1HUDOnly 26(AH) H6 No HUD Practice NoHUD Practice No HUD Practice NoHUD Practice No HUD Practice NoHUD H7 Practice H4 H5 25(VA) 36(AH) 47(HV) 58(VAH) 34(H) 38(VAH ) 33(A) 35(VA) 48(VAH) 45(VA) 53(A) 57(HV) 42(V) 48(VAH) 58(VAH) 56(AH) 1HUDOnly 23(A) 44(H) 52(V) 1HUDOnly 46(AH) 55(VA) 1HUDOnly 24(H) 1HUDOnly 27(HV) 38(VAH ) 28(VAH) 37(HV) 1HUDOnly 28(VAH) 32(V) 1HUDOnly 25(VA) 36(AH) 43(A) 48(VAH) 54(H) 57(HV) H10 No HUD Practice NoHUD Practice No HUD Practice NoHUD 1HUDOnly 22(V) 43(A) 54(H) H11 Practice 1HUDOnly 45(VA) 58(VAH) 1HUDOnly 28(VAH) 34(H) 1HUDOnly 27(HV) 38(VAH ) 1HUDOnly 23(A) 32(V) 1HUDOnly 28(VAH) 35(VA) 42(V) 46(AH) 53(A) 55(VA) 44(H) 47(HV) 58(VAH) 56(AH) 1HUDOnly 24(H) 48(VAH) 52(V) H8 H9 H12 H13 H14 H15 H16 No HUD Practice NoHUD Practice No HUD Practice NoHUD Practice No HUD Practice NoHUD 26(AH) 38(VAH ) 37(HV) 33(A) Table F4: Order in which conditions were presented to Highway Patrol Officers. 6 Table F5 below indicates the randomization of the tests performed by the ambulance drivers in the second driver simulator experiment. Subject A1 Practice NoHUD A2 Practice No HUD A3 Practice NoHUD A4 Practice No HUD A5 Practice NoHUD A6 Practice No HUD A7 Practice NoHUD A8 Practice No HUD A9 Practice NoHUD A10 Practice No HUD A11 Practice NoHUD A12 Practice No HUD A13 Practice NoHUD A14 Practice No HUD 1HUDOnly 24(H) 32(V) 1HUDOnly 28(VAH) 36(AH) 43(A) 47(HV) 1HUDOnly 23(A) 1HUDOnly 26(AH) 48(VAH) 52(V) 48(VAH) 57(HV) 34(H) 35(VA) 58(VAH) 55(VA) 1HUDOnly 28(VAH) 33(A) 42(V) 1HUDOnly 27(HV) 38(VAH) 45(VA) 54 56(AH) 1HUDOnly 22(V) 1HUDOnly 25(VA) 53(A) 58(VAH) 38(VAH) 44(H) 37(HV) 46(AH) 1HUDOnly 24(H) 33(A) 1HUDOnly 28(VAH) 35(VA) 1HUDOnly 23(A) 1HUDOnly 27(HV) 48(VAH) 52(V) 47(HV) 56(AH) 32(V) 44(H) 38(VAH) 46(AH) 1HUDOnly 28(VAH) 34(H) 1HUDOnly 26(AH) 37(HV) 42(V) 45(VA) 58(VAH) 55(VA) 53(A) 58(VAH) Table F5: Order in which conditions were presented to Ambulance Drivers. 7 Appendix G Human Factors Instructions and Questionnaires to test subjects for Experiments One and Two. Main Instructions (Simulator Experiment 1) Before you drive in our driving simulator today, we would like you to listen to these instructions. After you’ve heard them, we will answer any questions you have. Then, we will ask you to take a series of short drives in simulated conditions of poor visibility. This study is part of a series in which we are assessing new automobile technologies. These technologies aim to help drivers by providing useful information and warnings. The study is part of the Trunk Highway 7 Project, which has the overall objective of increasing safety on Highway 7 between Minneapolis and Hutchinson. We are particularly concerned with driving in bad weather, or poor visibility conditions, when it is difficult to see the road markings & when it is hard to see other vehicles ahead—particularly those that are parked or have been abandoned. Recently, new technologies have been developed that make it is possible—even in bad weather—to pick up lane markings and locate vehicles and other objects ahead of the driver. We will display this information to the driver. What we want to do in this experiment is to find out whether drivers can use this kind of information, and we want to determine whether it makes it easier for them to drive in poor visibility conditions. This car has been equipped with a head-up display. The display will show you where the road is and where any vehicles ahead might be—so that, even when the visibility is poor and it is difficult to see the lane markings or other vehicles, you will still be able to drive. On the display you will see two things. First, you will see the edge- and centerline markings of the road ahead. These projected markings will overlay the actual edge- and center-lines on the road ahead. And second, if there is another vehicle ahead of you, that is within 350 ft, you will see a rectangular outline that highlights the actual vehicle on the road. As well as continuously showing the road and letting you know about other vehicles, the head-up display may also give you two kinds of warnings—they are lane departure warnings and collision avoidance warnings. You may see a lane departure warning, if you are going out of lane, without meaning to. And you 1 may see a collision avoidance warning if there is a vehicle ahead that is stationary or going so slow that you could run into it. In this experiment, we are testing different ways of giving you these warnings. As I already mentioned, when you are in the simulator, you will drive several times—and before each drive, I will tell about the warnings and how they will be presented to you. But, first, do you have any questions? 2 Instructions for Specific Trials (Simulator Experiment 1) None & None For this drive, there will be no lane departure warnings. And no collision avoidance warnings. None & Red Only For this drive, there will be no lane departure warnings. But, you will be given a collision avoidance warning—if there is a vehicle ahead of you that is stationary, or going so slowly that you could run into it. When you get very close to that vehicle, you will receive the warning. The outline of the vehicle ahead will change color—it will change to red. If you get a red collision avoidance warning you should take appropriate action—either slowing down or steering around the vehicle. None & Yellow & Red For this drive, there will be no lane departure warnings. But, you will be given a collision avoidance warning and an advisory—if there is a vehicle ahead of you that is stationary, or going so slowly that you could run into it. The advisory will come first. When you are getting close to the vehicle ahead, the outline of that vehicle will change color—it will change to yellow. If you get this yellow advisory, you should be ready to take appropriate action. You will continue traveling toward the vehicle ahead and, when you are very close, you will receive a collision avoidance warning. The outline of the vehicle ahead will change color again—this time it will go red. If you get a red collision avoidance warning you should take appropriate action—either slowing down or steering around the vehicle. Red Area & None For this drive, you will be given a lane departure warning if you start to go out of lane. You will be given the warning if your front wheel touches the lane marker, either to the left or right. If your wheel does touch the lane marker, the area next to that lane marker will change color—it will change to red. And, if you get this red area warning, you should steer back into lane. For this drive, there will be no collision avoidance warnings. Red Area & Red Only 3 For this drive, you will be given a lane departure warning if you start to go out of lane. You will be given the warning if your front wheel touches the lane marker, either to the left or right. If your wheel does touch the lane marker, the area next to that lane marker will change color—it will change to red. If you get this red area warning, you should steer back into lane. Also for this drive, you will be given a collision avoidance warning—if there is a vehicle ahead of you that is stationary, or going so slowly that you could run into it. When you get very close to the vehicle ahead, you will receive a collision avoidance warning. The outline of the vehicle ahead will change color—it will change to red. If you get a red collision avoidance warning you should take appropriate action—either slowing down or steering around the vehicle. Red Area & Yellow and Red For this drive, you will be given a lane departure warning if you start to go out of lane. You will be given the warning if your front wheel touches the lane marker, either to the left or right. If your wheel does touch the lane marker, the area next to that lane marker will change color—it will change to red. If you get this red area warning, you should steer back into lane. Also for this drive, you will be given a collision avoidance warning and an advisory—if there is a vehicle ahead of you that is stationary, or going so slowly that you could run into it. The advisory will come first. When you are getting close to the vehicle ahead, the outline of that vehicle will change color—it will change to yellow. If you get this yellow advisory, you should be ready to take appropriate action. You will continue traveling toward the vehicle ahead and, when you are very close, you will receive a collision avoidance warning. The outline of the vehicle ahead will change color again—this time it will go red. If you get a red collision avoidance warning you should take appropriate action—either slowing down or steering around the vehicle. Area & None For this drive, you will be given a lane departure warning if you start to go out of lane. You will be given the warning if your front wheel touches the lane marker, either to the left or right. If your wheel does touch the lane marker, the area next to that lane marker will be shaded—it will be shaded the same color as the lane marker. If you get this shaded area warning, you should steer back into lane. For this drive, there will be no collision avoidance warnings. Area & Red Only 4 For this drive, you will be given a lane departure warning if you start to go out of lane. You will be given the warning if your front wheel touches the lane marker, either to the left or right. If your wheel does touch the lane marker, the area next to that lane marker will be shaded—it will be shaded the same color as the lane marker. If you get this shaded area warning, you should steer back into lane. Also for this drive, you will be given a collision avoidance warning—if there is a vehicle ahead of you that is stationary, or going so slowly that you could run into it. When you get very close to the vehicle ahead, you will receive a collision avoidance warning. The outline of the vehicle ahead will change color—it will change to red. If you get a red collision avoidance warning you should take appropriate action—either slowing down or steering around the vehicle. Area & Yellow and Red For this drive, you will be given a lane departure warning if you start to go out of lane. You will be given the warning if your front wheel touches the lane marker, either to the left or right. If your wheel does touch the lane marker, the area next to that lane marker will be shaded—it will be shaded the same color as the lane marker. If you get this shaded area warning, you should steer back into lane. Also for this drive, you will be given a collision avoidance warning and an advisory—if there is a vehicle ahead of you that is stationary, or going so slowly that you could run into it. The advisory will come first. When you are getting close to the vehicle ahead, the outline of that vehicle will change color—it will change to yellow. If you get this yellow advisory, you should be ready to take appropriate action. You will continue traveling toward the vehicle ahead and, when you are very close, you will receive a collision avoidance warning. The outline of the vehicle ahead will change color again—this time it will go red. If you get a red collision avoidance warning you should take appropriate action—either slowing down or steering around the vehicle. Red Line & None For this drive, you will be given a lane departure warning if you start to go out of lane. You will be given the warning if your front wheel touches the lane marker, either to the left or right. If your wheel does touch the lane marker, that lane marker will change color—it will change to red. If you get this red marker warning, you should steer back into lane. For this drive, there will be no collision avoidance warnings. 5 Red Line & Red Only For this drive, you will be given a lane departure warning if you start to go out of lane. You will be given the warning if your front wheel touches the lane marker, either to the left or right. If your wheel does touch the lane marker, that lane marker will change color—it will change to red. If you get this red marker warning, you should steer back into lane. Also for this drive, you will be given a collision avoidance warning—if there is a vehicle ahead of you that is stationary, or going so slowly that you could run into it. When you get very close to the vehicle ahead, you will receive a collision avoidance warning. The outline of the vehicle ahead will change color—it will change to red. If you get a red collision avoidance warning you should take appropriate action—either slowing down or steering around the vehicle. Red Line & Yellow & Red For this drive, you will be given a lane departure warning if you start to go out of lane. You will be given the warning if your front wheel touches the lane marker, either to the left or right. If your wheel does touch the lane marker, that lane marker will change color—it will change to red. If you get this red marker warning, you should steer back into lane. Also for this drive, you will be given a collision avoidance warning and an advisory—if there is a vehicle ahead of you that is stationary, or going so slowly that you could run into it. The advisory will come first. When you are getting close to the vehicle ahead, the outline of that vehicle will change color—it will change to yellow. If you get this yellow advisory, you should be ready to take appropriate action. You will continue traveling toward the vehicle ahead and, when you are very close, you will receive a collision avoidance warning. The outline of the vehicle ahead will change color again—this time it will go red. If you get a red collision avoidance warning you should take appropriate action—either slowing down or steering around the vehicle. Double Line & None For this drive, you will be given a lane departure warning if you start to go out of lane. You will be given the warning if your front wheel touches the lane marker, either to the left or right. If your wheel does touch a lane marker, that lane marker will change—it will change to a double line. If you get this double-line warning, you should steer back into lane. For this drive, there will be no collision avoidance warnings. 6 Double Line & Red Only For this drive, you will be given a lane departure warning if you start to go out of lane. You will be given the warning if your front wheel touches the lane marker, either to the left or right. If your wheel does touch a lane marker, that lane marker will change—it will change to a double line. If you get this double-line warning, you should steer back into lane. Also for this drive, you will be given a collision avoidance warning—if there is a vehicle ahead of you that is stationary, or going so slowly that you could run into it. When you get very close to the vehicle ahead, you will receive a collision avoidance warning. The outline of the vehicle ahead will change color—it will change to red. If you get a red collision avoidance warning you should take appropriate action—either slowing down or steering around the vehicle. Double Line & Yellow & Red For this drive, you will be given a lane departure warning if you start to go out of lane. You will be given the warning if your front wheel touches the lane marker, either to the left or right. If your wheel does touch a lane marker, that lane marker will change—it will change to a double line. If you get this double-line warning, you should steer back into lane. Also for this drive, you will be given a collision avoidance warning and an advisory—if there is a vehicle ahead of you that is stationary, or going so slowly that you could run into it. The advisory will come first. When you are getting close to the vehicle ahead, the outline of that vehicle will change color—it will change to yellow. If you get this yellow advisory, you should be ready to take appropriate action. You will continue traveling toward the vehicle ahead and, when you are very close, you will receive a collision avoidance warning. The outline of the vehicle ahead will change color again—this time it will go red. If you get a red collision avoidance warning you should take appropriate action—either slowing down or steering around the vehicle. Main Instructions (Simulator Experiment 2) Before you drive in the driving simulator today, we would like you to listen to these instructions. After you’ve heard them, we will answer any questions you have. Then, we will ask you to take a series of short drives in which there will be simulated snow cover and poor visibility. 7 This study is part of a series in which we are assessing new automobile technologies. These technologies aim to help drivers by providing useful information and warnings. The study is part of the Trunk Highway 7 Project, which has the overall objective of increasing safety on Highway 7 between Minneapolis and Hutchinson. We are particularly concerned with driving in bad weather, when it is very difficult— sometimes impossible—to see the road markings, & when it is hard to see other vehicles ahead—particularly those that are parked or have been abandoned. Recently, new technologies have been developed that make it possible—even in very bad weather—to pick up lane markings and locate vehicles and other objects ahead of the driver. What we want to do in this experiment is to find out whether you as drivers can use this kind of information, and we want to determine whether it makes it easier for you to drive in poor visibility conditions. This car has been equipped with a head-up display. There is a head-up display in front of you. It will show you where the road is and where any vehicles ahead might be—so that, even when the visibility is very bad and it is difficult to see the lane markings or other vehicles, you will still be able to drive. On the display you will see two things. First, you will see the edge- and center-line markings of the road ahead. These projected markings will overlay the actual edge- and center-lines on the road ahead. And second, if there is another vehicle ahead of you— that is within 350 ft—you will see a rectangular outline that highlights the actual vehicle on the road. The head-up display will also give you collision avoidance warnings. You will see a collision avoidance warning—if there is a vehicle ahead of you that is stationary, or going so slowly that you could run into it. When you get very close to that vehicle, you will receive the warning. The outline of the vehicle ahead will change color—it will change to red. If you get a red collision avoidance warning you should take appropriate action—either by slowing down or by steering around the vehicle. So, the head up display will continuously show you the road and let you know about other vehicles, and will give you collision avoidance warnings. Also, in some drives you will be given a lane departure warning, if you are going out of lane, without meaning to. We are testing different ways of giving this warning. It may be given to you on the head-up display. In this case the lane marker on the HUD will change color—it will change to red. Or you may hear the sound of a rumble strip. Or you may feel a vibration like a rumble strip in the side of the driver’s seat. Or you may get some combination of these warnings. Today, you will drive several times—and before each drive, I will tell how the lane departure warning will be given to you. 8 Specific Instructions (Simulator Experiment 2) Visual/Red Line For this drive, if you start to go out of lane, you will be given lane departure warnings that you see on the head-up display. You will be given a warning if your front wheel touches a lane marker. If your wheel does touch a lane marker, then on the HUD a lane marker will change color—it will change to red. If you are going out of lane to the right, the warning will be given on the right; and if you are going out of lane to the left, the warning will be given on the left. If you see a red marker warning, you should steer back into lane. Auditory/Rumble Sound For this drive, if you start to go out of lane, you will be given lane departure warnings that you hear. You will be given a warning if your front wheel touches a lane marker. If your wheel does touch a lane marker, then you will hear a warning—it will sound as if you are going over a rumble strip. If you are going out of lane to the right, the three warnings will be given on the right; and if you are going out of lane to the left, the three warnings will be given on the left. If you hear a rumble-strip warning, you should steer back into lane. Active seat For this drive, if you start to go out of lane, you will be given lane departure warnings that you feel in the driving seat. You will be given a warning if your front wheel touches a lane marker. If your wheel does touch a lane marker, then you will feel a warning in the driver’s seat—there will be a vibration that feels like going over a rumble strip. If you are going out of lane to the right, the warning will be given on the right; and if you are going out of lane to the left, the warnings will be given on the left. If you feel a rumble-strip warning, you should steer back into lane. Visual/Red Line & Auditory/Rumble Sound For this drive, if you start to go out of lane, you will be given lane departure warnings in two ways—you will see and hear the warnings. You will be given a warning if your front wheel touches a lane marker. If your wheel does touch a lane marker, then on the HUD a lane marker will change color—it will change to red. At the same time, you will hear a warning—it will sound as if you are going over a rumble strip. If you are going out of lane to the right, the two warnings will be given on the right; and if you are going out of lane to the left, the three warnings will be given on the left. If you see a red marker warning or hear a rumble-strip warning, you should steer back into lane. 9 Auditory/Rumble Sound & (Haptic) Driving Seat For this drive, if you start to go out of lane, you will be given lane departure warnings in two ways—you will hear and feel the warnings. You will be given a warning if your front wheel touches a lane marker. If your wheel does touch a lane marker, then you will hear a warning—it will sound as if you are going over a rumble strip. At the same time, you will feel a warning in the driver’s seat—there will be a vibration that feels like going over a rumble strip. If you are going out of lane to the right, the two warnings will be given on the right; and if you are going out of lane to the left, the two warnings will be given on the left. If you hear a rumble-strip warning or feel a rumble-strip warning, you should steer back into lane. (Haptic) Driving Seat & Visual/Red Line For this drive, if you start to go out of lane, you will be given lane departure warnings in two ways—you will hear and feel the warnings. You will be given a warning if your front wheel touches a lane marker. If your wheel does touch a lane marker, then you will feel a warning in the driver’s seat—there will be a vibration that feels like going over a rumble strip. At the same time, on the HUD a lane marker will change color—it will change to red. If you are going out of lane to the right, the two warnings will be given on the right; and if you are going out of lane to the left, the two warnings will be given on the left. If you feel a rumble-strip warning, or see a red marker warning, you should steer back into lane. Visual/Red Line & Auditory/Rumble Sound & (Haptic) Driving Seat For this drive, if you start to go out of lane, you will be given lane departure warnings in three ways—you will see, and hear, and feel the warnings. You will be given a warning if your front wheel touches a lane marker. If your wheel does touch a lane marker, then on the HUD a lane marker will change color—it will change to red. At the same time, you will hear a warning—it will sound as if you are going over a rumble strip. And you will feel a warning in the driver’s seat—there will be a vibration that feels like going over a rumble strip. If you are going out of lane to the right, the three warnings will be given on the right; and if you are going out of lane to the left, the three warnings will be given on the left. If you see a red marker warning, or hear a rumble-strip warning, or feel a rumble-strip warning, you should steer back into lane. 10 Questionnaire Form Single Modality (Experiment 2) Questions The following questions deal with your experience of driving in the simulator today. Each question is followed by a scale. This scale allows for a range of possible answers— from extremely favorable to extremely unfavorable. For each question, we would like you to make a mark on the scale in the place that best indicates how you feel about that question. For example: If you were asked, “how would you rate the importance of air bags in driver safety?” you might answer as shown below— Your answer Completely Unnecessary Absolutely necessary There are two sets of questions. The first set deals with the head-up display—but not with the lane departure warnings. And the second set will deal with lane departure warnings. Below you will find a set of four questions about the usefulness of the head-up display and the information it provided you today. 1. When you drove in poor visibility conditions today, how useful were the simulated lane markings provided by the head-up display? Not at all useful Very useful 11 2. When you drove in poor visibility conditions today, how useful was it for the head-up display to let you know that there were vehicles ahead? Not at all useful Very useful 3. When you drove in poor visibility conditions today, how useful were the collision avoidance warnings provided by the head-up display? Not at all useful Very useful 4. How easy was it for you to drive in the poor visibility conditions you experienced in the simulator while you were using the head-up display? Very difficult Very easy The following questions deal with lane departure warnings. 5. When you drove in poor visibility conditions today, how useful were the lane departure warnings? Not at all useful Very useful 12 6. When you drove in poor visibility conditions today, how useful was the red line (on the head-up display) for letting you know you were out of lane? Not at all useful Very useful 7. When you drove in poor visibility conditions today, how useful was the rumble-strip sound for letting you know you were out of lane? Not at all useful Very useful 8. When you drove in poor visibility conditions today, how useful was the vibration (in the driver's seat) for letting you know you were out of lane? Not at all useful Very useful 9. When you drove in poor visibility conditions today, how useful were the red line (on the head-up display) and rumble strip sound and vibration (in the driver's seat) for letting you know you were out of lane? Not at all useful Very useful 13 Finally do you have any comments about your experience of driving in the simulator today? 14 15 Questionnaire Form Dual Modality (Experiment 2) Questions The following questions deal with your experience of driving in the simulator today. Each question is followed by a scale. This scale allows for a range of possible answers— from extremely favorable to extremely unfavorable. For each question, we would like you to make a mark on the scale in the place that best indicates how you feel about that question. For example: If you were asked, “How would you rate the importance of air bags in driver safety?” you might answer as shown below— Your answer Completely unnecessary Absolutely necessary There are two sets of questions. The first set deals with the head-up display—but not with the lane departure warnings. And the second set will deal with lane departure warnings. Below you will find a set of four questions about the usefulness of the head-up display and the information it provided you today. 5. When you drove in poor visibility conditions today, how useful were the simulated lane markings provided by the head-up display? Not at all useful Very useful 16 6. When you drove in poor visibility conditions today, how useful was it for the head-up display to let you know that there were vehicles ahead? Not at all useful Very useful 7. When you drove in poor visibility conditions today, how useful were the collision avoidance warnings provided by the head-up display? Not at all useful Very useful 8. How easy was it for you to drive in the poor visibility conditions you experienced in the simulator while you were using the head-up display? Very difficult Very easy 17 The following two questions deal with lane departure warnings. 5. When you drove in poor visibility conditions today, how useful were the lane departure warnings? Not at all useful Very useful Finally, do you have any comments about your experience of driving in the simulator today? 18 19 APPENDIX H ANOVA Summary Tables for Experiment 2 Lane Departure Duration ANOVA Summaries The ANOVA summary tables shown below are the computer-generated summaries of the analyses of variance conducted on the lane-departure duration data obtained for the snowplow operators, highway patrol officers and ambulance drivers. There are two summaries for each type of specialty vehicle officer—the first is for those who had the three single-modality and the triple modality lane-departure warnings; the second is for those who had the three dual-modality and the triple modality lane-departure warnings. After the two ANOVA summaries for each type of specialty vehicle officer—presented in Table H1 through Table H6—there are two more ANOVA summary tables. For these two ANOVAs, the data from three types of specialty vehicle officer were combined. Table H7 combines the data from all the specialty vehicle operators who received the single-modality lane-departure warnings (as well as the triple-modality combination). Similarly, Table H8 combines the data from all the specialty vehicle operators who received three dual-modality warnings (and the triple modality combination). Tests of Between-Subjects Effectsb Dependent Variable: LOGDUR Source Corrected Model Intercept SUBJECT WARNING SUBJECT * WARNING Error Total Corrected Total Type III Sum of Squares 7.320a 1.631 2.596 .348 2.795 9.043 20.114 16.363 df 41 1 12 3 26 96 138 137 Mean Square .179 1.631 .216 .116 .107 9.420E-02 F 1.895 17.312 2.297 1.230 1.141 Sig. .006 .000 .013 .303 .314 a. R Squared = .447 (Adjusted R Squared = .211) b. Driver Type = Snowplow, Single or Dual Modality = Single Table H1: Full summary of ANOVA conducted for Snowplow Drivers who had the three single-modality and the triple modality lane-departure warnings. 1 Tests of Between-Subjects Effectsb Dependent Variable: LOGDUR Source Corrected Model Intercept SUBJECT WARNING SUBJECT * WARNING Error Total Corrected Total Type III Sum of Squares 3.293a 1.561 1.000 9.835E-02 2.247 4.291 11.373 7.584 df 38 1 11 3 24 64 103 102 Mean Square 8.666E-02 1.561 9.088E-02 3.278E-02 9.361E-02 6.705E-02 F 1.293 23.284 1.355 .489 1.396 Sig. .180 .000 .216 .691 .146 a. R Squared = .434 (Adjusted R Squared = .098) b. Driver Type = Snowplow, Single or Dual Modality = dual Table H2: Full summary of ANOVA conducted for Snowplow Drivers who had the three dual-modality and the triple modality lane-departure warnings. [Note, one participant in this group had no lane departures in any of the four trials and is, therefore omitted from this analysis.] Tests of Between-Subjects Effectsb Dependent Variable: LOGDUR Source Corrected Model Intercept SUBJECT WARNING SUBJECT * WARNING Error Total Corrected Total Type III Sum of Squares 5.751a .138 4.463 .455 .622 4.626 15.138 10.378 df 17 1 6 3 8 62 80 79 Mean Square .338 .138 .744 .152 7.775E-02 7.462E-02 F 4.534 1.845 9.970 2.032 1.042 Sig. .000 .179 .000 .119 .415 a. R Squared = .554 (Adjusted R Squared = .432) b. Driver Type = Highway Patrol, Single or Dual Modality = Single Table H3: Full summary of ANOVA conducted for Highway Patrol Officers who had the three single-modality and the triple modality lane-departure warnings. [Note, there was a data collection error for one participant in this group; because of this, no lane-departure data was available for this participant who had to be omitted from this analysis.] 2 Tests of Between-Subjects Effectsb Dependent Variable: LOGDUR Source Corrected Model Intercept SUBJECT WARNING SUBJECT * WARNING Error Total Corrected Total Type III Sum of Squares 2.127a 6.227E-02 .811 .797 .650 1.772 3.986 3.899 df 24 1 7 3 14 20 45 44 Mean Square 8.861E-02 6.227E-02 .116 .266 4.641E-02 8.860E-02 F 1.000 .703 1.308 2.997 .524 Sig. .505 .412 .297 .055 .891 a. R Squared = .545 (Adjusted R Squared = .000) b. Driver Type = Highway Patrol, Single or Dual Modality = dual Table H4: Full summary of ANOVA conducted for Highway Patrol Officers who had the three dual-modality and the triple modality lane-departure warnings. Tests of Between-Subjects Effectsb Dependent Variable: LOGDUR Source Corrected Model Intercept SUBJECT WARNING SUBJECT * WARNING Error Total Corrected Total Type III Sum of Squares .667a .184 .184 .126 .331 3.229 4.031 3.896 df 16 1 5 3 8 50 67 66 Mean Square 4.168E-02 .184 3.676E-02 4.216E-02 4.137E-02 6.458E-02 F .645 2.842 .569 .653 .641 Sig. .831 .098 .723 .585 .740 a. R Squared = .171 (Adjusted R Squared = -.094) b. Driver Type = Ambulance, Single or Dual Modality = Single Table H5: Full summary of ANOVA conducted for Ambulance Drivers who had the three single-modality and the triple modality lane-departure warnings. [Note, one participant in this group had no lane departures in any of the four trials and is, therefore omitted from this analysis.] 3 Tests of Between-Subjects Effectsb Dependent Variable: LOGDUR Source Corrected Model Intercept SUBJECT WARNING SUBJECT * WARNING Error Total Corrected Total Type III Sum of Squares 3.074a .145 .945 .741 .851 3.824 6.904 6.898 df 18 1 5 3 10 35 54 53 Mean Square .171 .145 .189 .247 8.510E-02 .109 F 1.563 1.329 1.730 2.262 .779 Sig. .126 .257 .154 .098 .648 a. R Squared = .446 (Adjusted R Squared = .160) b. Driver Type = Ambulance, Single or Dual Modality = dual Table H6: Full summary of ANOVA conducted for Ambulance Drivers who had the three dual-modality and the triple modality lane-departure warnings. Table H7: Full summary of ANOVA conducted for all three types of specialty vehicle operator who had the three single-modality and the triple modality lanedeparture warnings. 4 Table H8: Full summary of ANOVA conducted for all three types of specialty vehicle operator who had the three dual-modality and the triple modality lanedeparture warnings. 5 Appendix I Main Instructions (Rosemount Field Study) Before we start today, I would like you to listen to this background information. This study is part of a series in which we are assessing new automobile technologies. These technologies aim to help drivers by providing useful information and warnings. The study is part of the Trunk Highway 7 Project, which has the overall objective of increasing safety on Highway 7 between Minneapolis and Hutchinson. We are particularly concerned with driving in bad weather, when it is very difficult to see the road markings and other vehicles ahead—particularly those that are parked or have been abandoned. Recently, new technologies have been developed that make it possible—even in very bad weather—to pick up lane markings and locate vehicles and other objects ahead of you. This snowplow has been equipped with the head-up display that you see in front of you. This display will show you where the road is and can show you vehicles ahead—so that, when the visibility is very bad you will still be able to drive. Today, we would like you to drive in extremely poor visibility conditions using this display, so that you let us know whether it makes it easier for you to drive and whether there are ways in which we might improve it. On the display you will see right and left lane markings. These projected lane markings are centered in the road ahead. All three road segments that you drive on will be treated as one-way one-lane roads with the lane in the center of the road. In case of problems I need to mention two more things. First, when the head-up display is working, you will see a green bar on the display—it is in the lower right portion of the screen. When the display is working, this bar will flash—it will change from a solid to an outline, and then from an outline to a solid. If it stops flashing, it means the display is not working and that you should stop the snowplow. Second, if the screen should go red all over, it also means the display is not working and again you should stop the snowplow. If at any time you are uncomfortable driving, stop the snowplow and let me know. Do you have any questions?" After hearing the instructions the subject was informed that the lane departure warnings were visual, auditory, and haptic. She or he could see that there were no actual lane markings on the test track itself. However, the subject was told that if any part of the vehicle went outside the virtual lane markings all three warnings would trigger simultaneously--and on the same side that the lane departure occurred. If they departed to the right the warnings were given on the right side of the vehicle and if they departed to the left the warnings were given on the left side. They were asked if they had any questions. 1