Advanced Collision Avoidance Field Test Launched in Michigan
IVsource.net
29 March 2003

In one of the largest government-sponsored field trials of its kind, General Motors,  Delphi, and others will enlist Michigan drivers to test vehicles equipped with both forward collision warning and adaptive cruise control systems.   IVsource reports on their recent kickoff event at Milford Proving Grounds near GM Headquarters, describing the project's objectives and the team’s approach to some of the key technical challenges.

In one of the largest government-sponsored field trials of its kind, General Motors and a group of partners will enlist about 80 Michigan drivers to test vehicles equipped with both forward collision warning and adaptive cruise control systems.

The US Department of Transportation (DOT), GM, and Delphi Automotive are funding the project, formally called the Advanced Collision Avoidance System Field Operational Test (ACAS FOT).  The University of Michigan Transportation Research Institute (UMTRI) is under contract to the project to conduct the evaluation portion of the 10-month field test, beginning this month.  The test, involving 10 Buick LeSabres, is the latest phase of a five-year, $35-million partnership formed in 1999 to develop and evaluate collision avoidance technologies.  GM led the integration of the system and the assembly of the test vehicles.

GM and Delphi Delco Electronics provide the technical applications behind the adaptive cruise control (ACC), forward collision warning (FCW) and the driver interface.  The FCW system detects and informs drivers of  three types of rear-end crash scenarios: a tailgating advisory (triggered by following a preceding vehicle too closely), a cautionary closing alert, and an imminent closing alert.

First Integration of ACC and FCW

Although basic collision avoidance technologies are out in the marketplace, no one until now has developed multi-sensor technologies that integrate both adaptive cruise control and forward collision warning systems.  According to the project sponsors, the test system, developed by GM and Delphi, is an improvement over previous versions because of the system's ability to reduce false alerts and to detect more quickly a slowed or stopped vehicle around a curve.  

Furthermore, no one has ever performed such an extensive field test of these technologies.

"This program is the most extensive collision warning field operational test ever undertaken.  It is designed to gauge the effectiveness of such a system under real-world driving conditions, as well as driver acceptance of the technology," said Larry Burns, GM Vice President of Research & Development and Planning.  GM’s safety technology goal is to “fully realize the potential of technologies that allow us to avoid crashes,” Burns added.

UMTRI is managing the field test and analyzing the data.  Licensed men and women in their 20s, 40s and 60s from Southeastern Michigan were randomly selected with the assistance of the office of the Michigan Secretary of State.

"The research program is a cooperative agreement," said Ray Owings, Ph.D., the Associate Administrator at the National Highway Traffic Safety Administration (NHTSA) for Advanced Research and Analysis.   "That means [US]DOT has a responsibility to contribute technical information and provide other support to make the project a success."

USDOT has funded other projects related to adaptive cruise control and forward collision warning systems, and has shared the knowledge gained in those projects with this research program.  Additionally, NHTSA collaborated on the testing of the vehicles and provided comparative data.  "NHTSA is supporting research that will analyze the data collected during the field operational test," added Dr. Owings.  The USDOT’s Volpe Center will be conducting an independent evaluation of the data.

Researchers will study, among other things, if drivers using the systems actually experience fewer "close following" or "rapid-closing" driving situations that could lead to crashes, and if the performance of these systems meets consumer expectations.

How the System Works

The forward collision warning system being tested uses electronic sensors, Global Positioning System (GPS) technology, and radar to provide audio and visual warnings to a driver who is approaching a slowed or stopped object too rapidly, or who is following a vehicle too closely.  The warning signals to the driver that he or she may need to brake quickly or make an evasive maneuver to avoid a collision.  The visual warnings are illuminated in front of the driver on a head-up display on the windshield.

"The head-up display is a very important feature of these tests, and it is in keeping with GM's philosophy to maximize drivers' ability to keep their eyes on the road and their hands on the wheel," said Burns.

Adaptive cruise control greatly enhances the convenience of cruise control, especially in traffic.  The system uses the same sensors as the forward collision warning system, including the radar sensor mounted at the front of the car to detect objects in its path.  If the lane ahead is clear, the system will maintain the set speed, just like conventional cruise control.  When a vehicle is detected in the same lane in front of the car, the system will adjust vehicle speed by applying limited braking or acceleration to maintain a driver-selected 'follow distance' (or 'time headway') to the vehicle ahead.  This system also provides visual and auditory warning to drivers when they need to take over braking control, and provides forward collision warnings as well.

Sensor Fusion is Core to Vehicle Intelligence

At the press event, Dr. Ray Kiefer, ACAS Program Manager, and Dr. Jeremy Salinger, ACAS Technical Director, provided some insights into the technical challenges the team has face in implementing these functions.  In order to work reliably, the system must estimate where the road is going, where the driver is going, and determine the nearest vehicle in the projected path of the driver.  This type of estimation is particularly difficult when road curvature transitions from straight to curved, and vice versa.

Dr. Kiefer described the key new methods explored in ACAS.  First, a digital map and GPS receiver enable an indication of vehicle position and direction of travel on the map, which is used to predict road geometry ahead.  Second, a forward-looking machine-vision system uses lane markings to estimate the road geometry ahead of the vehicle.  And third, radar tracking uses the trajectories of tracked vehicles ahead to determine if there is a pattern that may indicate the upcoming road geometry (for instance, if all forward vehicles are slightly turning to the right on a highway, it is likely that the road itself is curving to the right).  Data fusion combines these estimates to determine the best overall prediction of road geometry ahead.

It’s All About Evaluation

Researchers say that the key to success is to understand how the driver uses the system and the system's overall impact on safety.   Over 500 data channels are being recorded in the ACAS test, including:

• Circumstances surrounding crash alert occurrences

• Roadway and face video

• Brake applications

• Vehicle speeds

• Traffic condition

• Driver-preferred system settings

Dr. Salinger noted a number of key issues being addressed:

1. (Primary)  Do drivers experience fewer tailgating or "approaching too fast" driving situations, which can lead to rear-end crashes?

2. Do drivers respond quickly and appropriately to visual and auditory warnings?

3. How often do drivers experience useful warnings vs. “cry wolf” false alarms?   Under what circumstances?

4. Do drivers have an accurate mental model of the system?

5. How do drivers feel about the crash alert timing and interface approach?

6. What ACC headway and FCW alert timing settings do drivers prefer?

7. Under that traffic conditions will drivers choose to use ACC?

8. What implications do these results have on customer education approaches for ACC and FCW systems?

9. Do drivers find ACC and FCW systems useful?

10. Are customer expectations  being met for ACC and FCW system performance (e.g. are drivers tolerant of false alarms?)

It will be the task of the government’s Volpe Center to sort through gigabytes of data to come to some conclusions on these questions.

Continuous Implementation of Advanced Technology

If ultimately incorporated into vehicles, forward collision warning and adaptive cruise control systems could help drivers avoid or reduce the number of rear-end crashes, which account for about 29 percent of all police-reported crashes.  (GM has separately announced that ACC has been introduced on the 2003 Cadillac XLR.  There are indications that the engineering development for this ACAS project flowed directly into the design and introduction of this product.)

GM’s “Advanced Technology Development Stream” through 2010 includes significantly more advanced features, such as lane keeping, intersection warning, and vehicle-to-vehicle communications for maneuver coordination.  (See related IVsource article.)

One of the 10 2002 Buick LeSabre test vehicles that will be used in the Forward Collision Warning/Adaptive Cruise Control field operational trial.

Automotive Crash Avoidance System and data acquisition systems are housed in the trunk of the test vehicles.

A close-up of the front-sensing radar device.

Examples of the driver interface displays.

[Top]

For More Information ...

... contact Mr. Jack Ference of NHTSA, federal project manager, at jference@nhtsa.dot.gov, or see www.its.dot.gov. 

[Top]