In the mid-1990s, GM's intelligent vehicle efforts focused on the automated highway system, which allowed novelties like drivers riding down the road while reading the newspaper - in perfect safety. But while impressive, the system demonstrated in 1997 was seen as ahead of its time in regard to the current commercial market. That's a major difference between the automated highway and GM's current Intelligent Vehicle Initiative research project, pursued along with the U.S. Department of Transportation and primary partner Delphi Delco Electronics.
The goal of the Automotive Collision Avoidance System (ACAS) is about as practical as can be: to help eliminate vehicle crashes. ACAS helps drivers avoid collisions by making the car smarter about the traffic and environment around it, detecting threats, and either controlling the vehicle through adaptive cruise control (ACC) or warning drivers with audible tones and visual alerts. The first type of crash targeted for elimination are rear-end collisions, which account for 28 percent of all injury accidents and frequently result from inattention, tailgating, or speeding.
The initial phase of the five-year, $35 million project involves development and testing of prototype vehicles outfitted with crash avoidance technology. There's little doubt that these Buick LeSabres will be better-equipped to avoid collisions than any other vehicle on the road. The system relies on four major elements:
- Forward Radar Sensor - Forward-looking radar detects and tracks obstacles and moving vehicles ahead, including those that may be temporarily obscured by another vehicle.
- Forward Vision System - A video system estimates the lane alignment, curvature or offset of the road up to 100 yards ahead. The system watches the relationship between the vehicle and lane and the angle the vehicle is headed.
- In-vehicle Sensors - Sensors monitor compass heading, lateral and longitudinal acceleration, yaw rate, steering angle and wheel speeds.
- Global Positioning System (GPS)/Map System - A GPS receiver and dead-reckoning navigation estimate the vehicle's position, then compare it to digital map databases. This gives the system a sense of the road ahead, where the car is on the road, and the key features along the way, such as hills, merging lanes, tunnels, or curves. This data also produces an estimate of the car's motion.
- In addition, ACAS keeps tabs on what the driver is doing to estimate his or her distraction level and response times.
"With all that information flooding in, one of the keys to operating
the system is deciding what's most believable," says
Ron
Colgin, Program Manager. "That's the job of the data fusion
algorithms, which assess each part of the data and integrate
them to produce the best picture of reality."
By tracking the host vehicle and those around it, the software singles out a target: the object the car would hit first if its motion did not change. At that point, the system conducts a threat assessment to determine if it needs to take action.
"We're trying to determine what the closest in-path vehicle is doing," explains Colgin, "but the vehicles in front of us don't always move on the same path our car does. They may be following a curved section while we're driving straight. So we're applying additional technology - GPS and the vision system - to determine what the road geometry ahead of us is. Knowing the road curvature beneath us is not enough."
When a threat is detected, ACAS will either alert the driver or, if adaptive cruise control is in use, take control of the accelerator and brakes to maintain a safe distance. The warning/alert features can be a combination of visual, auditory and haptic clues.
"If the driver's foot is on the brake or the car is moving less than 25 mph, ACAS will still be gathering and analyzing data but will not report any warnings," says Colgin. "That's because the sensors might report too much information in target-rich stop-and-go traffic, annoying the driver with constant warnings."
Determining how well the warning/alert system works will be part of the next phase of the project, which includes a 10-month field test in Southeastern Michigan. GM will build 10 production Buick LeSabres equipped with ACAS in 2002 for real-world testing. The field test will allow 120 drivers unrestricted use of a test car for two to four weeks while data is collected. The University of Michigan Transportation Research Institute will help screen, select and orient the drivers, then collect data and analyze the results. In the end, the project partners hope that the only crash they see is in the number of rear-end collisions.
By Diane Kightlinger
