|Type: Field operational test|
|Tested system/service: Autonomous Systems|
|Countries: USA||? test users|
|? partners||10 vehicles|
|Active from 06/1999 to 06/2004|
|Robert D. Ervin|
|University of Michigan Transportation Research Institute|
|Data catalogue||Tools catalogue|
|Data sets used in this FOT:
||The following tools|
were used in this FOT:
The Automotive Collision Avoidance System Field Operational Test (ACAS/FOT) was a cooperative agreement between General Motors and the U.S. Department of Transportation National Highway Transportation Administration. Under the agreement, General Motors and Delphi-Electronics Systems (DDE) worked as a team to refine and integrate the required technologies necessary to conduct the FOT. The U.S. DOT, General Motors, DDE and Delphi-Chassis Systems (DCS) have each provided funds for the execution of the project.
The ACAS FCW system was designed to provide visual and audible warnings to a driver if detected an imminent crash with the rear end of another vehicle. It also provided visual cues to help the driver maintain a safe distance when following other vehicles. The ACAS ACC system is a driving comfort and convenience feature that maintains a set speed when there is no impeding traffic, and that will reduce the speed to maintain a selected time headway when slower moving traffic impinges upon the path of the vehicle.
During the field operational test, lay drivers used vehicles equipped with the ACAS functions as their personal vehicles, unrestricted and unsupervised, for four weeks each. Extensive subjective and quantitative data were collected to help determine driver acceptance and the impact of the system on driving safety.
Details of Field Operational Test
Start date and duration of FOT execution
February - December 2003
The route originated in Ann Arbor, proceeded along mostly freeway segments into Southfield, went south on freeways and a 24-mile segment of Telegraph Road to Flat Rock, and proceeded back to Ann Arbor on mostly freeways. Each person who drove the route filled out a 21-point questionnaire and the results were compiled and summarized for informing later stages of the ACAS activity.
The goal of the research program was to demonstrate the state-of-the-art of rear-end collision warning systems and measure system performance and effectiveness using lay drivers on public roads in the United States.
The main mission of the ACAS/FOT Program was to identify key enabling technologies that can accelerate the development of a cohesive collision warning vehicle package which in turn can be used to assess the technological impact of a collision warning system through a comprehensive field operational test program. The performance of the cohesive collision warning vehicle package was of sufficient fidelity, robustness, and maturity so that a meaningful field operational test program can be executed.
In support of this mission, other secondary goals and objectives were also specified:
- Form a team that has demonstrated expertise and capability in the technology, manufacturing, and marketing of collision avoidance products.
- Leverage, capitalize, and exploit existing high-value developed portfolio of ACC and FCW technologies/component for implementation in the proposed ACAS/FOT Program. Of primary interest are the achieved successes from the initial ACAS Program and the recent development activities of other NHTSA/FHWA sponsored programs. These activities will provide value added program benefits by minimizing new learning curve experiences, preventing duplication of efforts, streamlining the system design process, and accelerating the activities of the proposed program.
- Incorporate human factors into the design process. Of primary interest are the successes achieved from the initial ACAS Program and the recent development activities of other NHTSA/FHWA sponsored programs of relevance.
- Utilize system engineering design procedures and practices to focus the accelerated development of a validated comprehensive collision warning system that is seamlessly upward integrated into the vehicle infrastructure. The tested and validated design will be used to produce a fleet of ten deployment vehicles for use in the field operational test program.
Summary of Key Forward Crash Warning (FCW) Results Pertaining to Driving Safety and Driver Acceptance
Results from analysis of the FCW-related data showed that:
- Driver response to the ACAS FCW system was mixed. Older drivers were more likely to view the system favorably, and middle-age drivers the least likely. Most drivers saw some limited benefit associated with the FCW system, but typically reported that the benefit would be greater for drivers other than themselves.
- After experiencing the FCW feature for three weeks, most of the FOT subjects were not willing to purchase such a system at a $1000 cost.
- The most important factor influencing the frequency and conditions in which individual drivers experienced alerts were the individual driver themselves, 16 with the type of road (and therefore, traffic dynamics) being the second most important factor.
- Drivers frequently commented that they received more FCW alerts than they believed were truly necessary, with the additional alerts being deemed as nuisances or false alarms. This seemed to contribute significantly to the negative perceptions of the FCW system.
- The two statements most frequently associated with FCW system attributes that needed improvement were first, reduce the frequency of nuisance and false alarms, and second, to provide a means to permit the FCW system to be turned off in certain types of traffic conditions.
- At least 13 situations were identified where FCW appeared to contribute to the driver’s proper awareness of a potential rear-end crash, and/or the encouragement of an appropriate, firm braking response to the situation.
- The headway distances during periods of vehicle-following in manual, daytime driving were also seen to increase on all road types (with no corresponding impact on nighttime driving.)
- No change in the rate or the severity of conflicts was observed when driving with versus without FCW.
- No consistent set of results suggested that driver braking responses to conflicts was either positively or negatively affected.
- A majority of FCW imminent alerts were either false alerts triggered by objects not on the roadway or alerts occurring in scenarios in which the forward conflict is typically resolved through a divergence in the paths of the two vehicles rather than through braking by the host driver. This aspect of system performance appears to have negatively influenced driver acceptance of FCW.
- The current state of sensor processing leaves FCW operating with much less information than an alert driver has regarding anticipated vehicle movements and the detection of vehicles that are stopped in one’s own path. Therefore, the most compelling change for improving the system, given the current state of the technology, would be the elimination of stationary-target alerts while still retaining the potential to warn on a ‘movable’ object that came to a stop in the vehicle’s path.
Results from analysis of the ACC-related data showed that:
- After experiencing the ACC feature for three weeks, most of the FOT subjects seemed genuinely willing to purchase such a system.
- The ability of this ACC controller to provide smooth, effective management of speed and headway over a very broad range of driving conditions is believed to account for its wide utilization and acceptance by FOT drivers.
- ACC driving was basically benign in all of its safety implications for freeway driving.
- The rather popular usage of ACC in dense, but flowing, freeway traffic does result in more cut-in activity ahead of the ACC vehicle due to the somewhat longer headway times that are managed by the system.
- The converse effect of longer headway-times, as well as the continuous control action of the ACC system is that ACC driving affords more headway clearance and lower levels of kinematic conflict on an ongoing basis.
- ACC can be kept continuously engaged over long distances on freeways, especially when traffic density is sparse.
- ACC driving on surface streets appears to pose a possible safety concern for the neophyte ACC user who will become exposed to the stronger conflicts that may arise in this environment.
- The fact that drivers adapted within only a three-week test window to significantly contain their exposure to conflict-laden driving conditions, such as surface streets, would seem to bode well for the long-term adoption of prudent practices of ACC supervision by the driver.
- All evidence indicates the FOT drivers managed the ACC system with a rather high state of attentiveness, especially as reflected in short driver braking reaction times and modest levels of deceleration, when braking interventions did occur.
- Although the tested ACC system was capable of automatically decelerating at up to 0.3g, the deliberately-retarded delivery of this response by the ACC controller is believed to have been an effective characteristic in discouraging drivers from depending upon it.
Driver response to the ACAS FCW system was mixed. Older drivers were more likely to view the system favorably, and middle-age drivers the least likely. Most drivers saw some limited benefit associated with the FCW system, but typically reported that the 10-15 benefit would be greater for drivers other than themselves. After experiencing the FCW feature for three weeks, most of the FOT subjects were not willing to purchase such a system.
There was substantial variation in the frequency of, and conditions in which, individual drivers experienced alerts from the FCW system. The most important factor influencing this experience appears to be the individual driver, with the type of road (and therefore, traffic dynamic and roadside environment) being the second most important factor.
Drivers frequently commented that they received more FCW alerts than they believed were truly necessary, with the additional alerts being deemed as nuisances or false alarms. This seemed to contribute significantly to what were generally negative perceptions of the FCW system. The two statements most frequently associated with FCW system attributes that needed improvement were first, to reduce the frequency of nuisance and false alarms, and second, to provide a means for turning off the FCW system in certain types of traffic conditions.
Overall there was no change in the rate or the severity of approach conflicts when driving with FCW versus driving without it. There was also no consistent set of results suggesting that driver braking responses to conflicts were either positively or negatively affected. The visual cautionary alerts appear to have introduced a short-lived change in headway following, but this effect is limited to middle-aged drivers. The headway distances during periods of vehicle-following in manual driving were also seen to increase on limited-access highways, as well as during daytime driving on all road types.
The source of these effects remains somewhat unclear, given the lack of a sustained effect due to cautionary alerts, but it may be that ACAS creates a general increase in drivers’ awareness of headway.
There were suggestions of safety potential of the FCW system. At least 13 situations were identified in which FCW appeared to contribute to the driver’s proper awareness of a potential rear-end crash, and/or an encouragement of appropriate firm braking response to the situation.
Regarding the state of maturity of the FCW system, a majority of FCW imminent alerts were either false alerts triggered by objects not on the roadway or alerts in which the driver anticipates the benign resolution of the conflict via lateral motions of one or both vehicles, with no braking required. These implied deficiencies of the system trace to the current state of sensors and sensory processing that leaves FCW operating with much less information than drivers have at their disposal, and therefore a lesser ability to identify relevant in-path objects and to predict the likely motions of the vehicles. This 10-16 aspect of system performance appears to have negatively influenced driver acceptance of FCW, and provides a primary area for system improvement.
The ACAS ACC system was widely used and favorably regarded by most participants. After experiencing the ACC feature for three weeks, most of the FOT subjects seemed genuinely willing to purchase such a system. The ability of this ACC controller to provide smooth, effective management of speed and headway over a very broad range of driving conditions is believed to account for its wide utilization by FOT drivers. A broadly distinctive feature that differentiates ACC control from any other driving mode is the reduced time spent at headway values below one second.
ACC was found to be basically benign in all of its safety implications for freeway driving. In particular, ACC driving produced fewer and lower-magnitude conflicts with other traffic on freeways, generally, as a result of either the system’s control performance or differences in the way the ACC driver tends to follow and pass other vehicles. The rather popular usage of ACC in dense, but flowing, freeway traffic does result in more cut-in activity ahead of the ACC vehicle due to the somewhat longer headway times that are managed by the system. Also, the very long distances that can be driven with ACC continuously engaged makes it quite possible to travel for hours without a braking intervention.
ACC driving on surface streets, at least during the first week or so of many driver’s exposure to the system, appears as an issue of some concern since the data showed that strong conflicts with other vehicles arise frequently in that environment. On the other hand, drivers adapted quickly to the task of supervising ACC such that the rate of full-auto braking incidents dropped precipitously within the short, three-week span of system testing.
Finally, although the tested ACC system was capable of automatically decelerating at up to 0.3g, the deliberately-retarded delivery of this response by the ACC controller is believed to have been an effective characteristic in discouraging drivers from depending upon it. Further, it appears that drivers were not generally able to experience the full 0.3-g braking response of ACC by means of experimentation.
Summary, type of funding and budget
U. S. Department of Transportation (Agreement No. DTNH22-99-H-07109)
Cooperation partners and contact persons
FOT conducted by the University of Michigan Transportation Research Institute (UMTRI), under a subcontract from the General Motors (GM). GM served as the prime contractor under a Cooperative Agreement with the U.S. Department of Transportation (USDOT) as part of its Intelligent Vehicle Initiative program. The Delphi Corporation served as another subcontractor to GM.
- Public Authorities:
- Vehicle Manufacturer:
- Research Institutes:
- Others (specify):
Applications and equipment
- Forward Radar Sensor
- Forward Vision Sensor
- Brake Control System
- Throttle Control System
- Driver-Vehicle Interface
- Data Fusion
- Tracking and Identification
- Collision Warning Function
- Adaptive Cruise Control Function
The FOT was conducted with a fleet of ten vehicles (plus a backup vehicle). These were drawn from the 13 vehicles fabricated by GM with Delphi’s assistance. The vehicle platform selected by GM was the Buick LeSabre, model year 2002.
These vehicles were virtually identical, having the same production options including a silver-metallic paint. On the vehicle’s outside, there were very few items that differentiated the test vehicle from ordinary LeSabres, so that these cars did not particularly stand out in traffic: FOT vehicles had visible antennas for GPS and the data acquisition system’s cellular modem, plus the radar fascia positioned in the center of the grill.
Inside, the vehicle differed from the stock LeSabre as well:
- A head-up display (HUD) was installed in front of the driver to project ACAS visual displays and vehicle speed within the driver’s field of view. Extra driver control switches were installed in the dashboard to the left of the steering column to control HUD brightness and vertical position.
- Existing buttons, located on the steering wheel, that were used for HVAC and radio functions were re-wired for use by the ACAS system. These allowed drivers to set the ACC gap (distance) and adjust the sensitivity (lateness) of the FCW visual cautionary alerts.
- A driver-face camera was mounted on the A-pillar.
- Two forward-looking cameras were installed near the top of the windshield on the passenger’s side of the rear-view mirror. A shroud shielded them from view of the passengers.
- A “comment button” was installed on the dash to the right of the steering wheel column, so that drivers could dictate comments on specific ACAS experiences when they so desired.
The driver would also note that the forward portion of the LeSabre’s trunk was closed off by a panel. Behind this panel were most of the ACAS processors and power-management systems. Despite these extra features, extensive care had been taken by GM and Delphi to minimize a driver’s sense that they were in a research vehicle.
The ACAS FOT was broadly successful both in terms of fielding a reliable test vehicle and recovering the desired data. Of the 96 individual drivers who were given an ACAS vehicle and launched as FOT subjects, all successfully completed the multi-week term of the driving assignment. In 13 cases, some problem with a deployed vehicle was resolved by substituting a replacement car for the faulty one in the field.
Equipment carried by test users
The video record was derived from forward-looking- and driver’s-face-oriented cameras. The forward scene was recorded on a continuous basis at 1 Hz while the driver’s face was sampled at 5 Hz for 4 seconds every 5 minutes. When an ACAS-alert event occurred, the forward and face-oriented cameras were recorded at 10Hz over an 8-second window that straddled the moment of alert-onset.
Pre-simulation / Piloting of the FOT
Method for the baseline
Techniques for measurement and data collection
Measured against the designed scope of the FOT, 94% of the intended data was successfully collected and compiled into a relational database. The resulting database of engineering variables is 164 GB in volume and contains up to 400 engineering variables that were sampled at 10-Hz, as well as subjective assessments. The companion set of video files is fully synchronized with the quantitative database.
Subjective data were collected using several questionnaires as well as an interactive debriefing that was done on the day the car was returned from the field by a participant.
The live debriefing involved a replay of approximately a dozen alert events, as seen through the forward- and face-looking cameras. Subjects were asked to rate each of the selected alerts, given their own recollection and observation of the driving circumstances in which the alert occurred. Later, four focus groups were held, each gathering several individuals for a structured discussion and evaluation of their ACAS-driving experience.
Recruitment goals and methods
Methods for the liaison with the drivers during the FOT execution
Methods for data analysis, evaluation, synthesis and conclusions
Sources of information
A comprehensive report with extensive data can be found at:
Final Report - Technical:
Final Report - Appendices: