May 4, 2004
Staying between the lines: Automotive innovations promise improved safety
By Geoff Koch
Snaking between lines of orange cones in a new Corvette, the last thing you would want to do is take your hands off the steering wheel. Unless, that is, you are Chris Gerdes, assistant professor of mechanical engineering and co-director of the Stanford Center for Design Research (CDR).
During a demonstration last week on the roof of a Stanford parking garage, Gerdes sat in the driver's seat, pedal to the metal but hands well clear of the steering wheel. A novel combination of global positioning system (GPS) and computer technologies, which learned the route after just one pass through the cones, did the rest.
"This technology might save tens of thousands of lives each year," Gerdes said.
Three GPS antennas on the car's roof identified the direction the car was traveling and its exact location with a few centimeters. Whenever the car started to drift, an onboard computer nudged it back toward the center of the lane.
Gerdes was clearly having fun, but his efforts to fine-tune "lane-keeping" technology are about more than building the ultimate remote- control car. The technology takes aim at some of the more grisly traffic safety statistics.
Motor vehicle accidents are the leading cause of death for people from childhood to their mid-30s, according to the Centers for Disease Control and Prevention. And in almost half of all fatal motor vehicle accidents, the first harmful event occurs when the vehicle leaves its lane and collides with a fixed object.
Most of the hardware installed in Gerdes' conspicuous and sporty prototype is of the standard, off-the-shelf variety. The gyroscopes that collect information about the pitch and yaw of the car cost just a few dollars each. The Pentium computer bolted into the car's trunk would be deemed slow and outdated in any computer cluster on the Stanford campus.
The real innovation from Gerdes' lab comes in the form of custom algorithms and models. His goal is to write software that creates an experience for the driver that is reliable yet unobtrusive.
When it comes to lane-keeping, the algorithm can be thought of as a spring with one end attached to the car and the other to the middle of the lane. If the car veers slightly, gentle tension in the "spring" pulls the car back in the opposite direction. If the veering is more pronounced, the spring responds with more force to pull the car back to the middle of its lane.
Drivers can override the technology anytime, as Gerdes showed during the parking garage test.
"If people don't like the way the technology works, they'll turn it off," Gerdes said, retaking control simply by grabbing the steering wheel.
Look, Ma, no hands!
Gerdes expects lane-keeping features to start showing up in cars within a decade. How quickly the technology arrives in showrooms will be determined in part by acceptance of its "drive-by-wire" system, he said.
In such a system the driver still uses a steering wheel, but it's connected to a computer instead of the steering column. This computer, in turn, communicates with the rest of car.
Work on lane-keeping technology is part of a broader effort that spans many universities and companies to improve safety and efficiency of cars through use of computers. Eventually, drivers may come to rely more on information technology than on roadmaps to get from point A to point B.
Stephen Dubowsky, professor of mechanical engineering at the Massachusetts Institute of Technology, is on sabbatical at Stanford and works just down the hall from Gerdes. "Traffic congestion could be reduced because cars could travel closer together at higher speeds," Dubowsky said of a future in which software, computer networks and sensor technology work in concert to optimize traffic flow.
A teacher who's driven
Though himself a graduate student less than 10 years ago, today Gerdes has a whopping 10 doctoral students under his wing. Some may be attracted by the chance to customize high-performance cars -- the lab has a similarly equipped Mercedes E320 sedan -- but all seem to appreciate his genuine enthusiasm for teaching and collaboration.
"He's created an environment where we can freely express our own opinions and ideas," said Paul Yih, a doctoral student from Pennsylvania doing research on drive-by-wire technology. "Yet he has taught us to maintain our focus on the things that really matter in research -- how does it benefit society, and how do we convince others that it does?"
Jean-Pierre Hathout, a Bosch scientist and program manager, recently collaborated with Gerdes in a separate project involving a new kind of clean-burning engine. "When I'd get a headache in my office, I'd just take my laptop and go work in [Gerdes'] lab," he said, "He has created a successful, winning atmosphere and it's clear his students love him."
The bulk of Gerdes' lane-keeping work is funded by the National Science Foundation, though his lab also has received support from Bosch, General Motors, DaimlerChrysler and Nissan. Would he ever be tempted to leave academia for a plush job at one of these corporate benefactors?
"No," said Gerdes, who worked for two years for Daimler-Benz North America and Freightliner before joining the Stanford faculty. "I just love teaching."
The black 'Vette is a nice perk, too.
Geoff Koch is a science-writing intern with the News Service.
This release was written by science-writing intern Geoff Koch. Photos are available on the Web at http://newsphotos.stanford.edu.