Defense Advanced Research Projects Agency photo
	Stanley, the Stanford University team's entry in last month's Grand Challenge Race in Nevada, crosses the finish line first. MORE COVERAGE Strategy, engine trouble factored in CMU vehicle's unexpected finish

"Herbie, meet Stanley."

That was the message of Volkswagen's recent full-page ad in USA Today, comparing the fictional Love Bug, a car with a mind of its own on the race track, with the real-life Stanley, a VW Touareg with a true mind of its own.

It was Stanford University computer scientists who gave Stanley the smarts to drive itself in the 132-mile Grand Challenge robot race on Oct. 8, but Volkswagen's corporate pride is understandable, as is the play on its advertising slogan -- "Drivers Wanted (but not required)."

Stanley might have taken home the $2 million prize awarded by the Defense Advanced Research Projects Agency, but four other robotic vehicles also completed the course through the Nevada desert without human intervention or guidance, outstripping the expectations of many observers.

"It's not a victory for a specific institution," said Sebastian Thrun, director of the Stanford Artificial Intelligence Lab and leader of the Stanford Racing Team. "It was a victory for the field."

Indeed, Dr. Thrun's telephone has been ringing regularly ever since, with proposals from defense contractors, entrepreneurs and others with ideas of how to employ autonomous navigation. But so has the phone of William "Red" Whittaker, the Carnegie Mellon University roboticist whose Red Team had the second- and third-place finishers.

"The thing we've got going for us is the success of the Grand Challenge shifted a lot of the belief state," Dr. Whittaker said last week. In just the past month, a number of government "requests for proposals" have appeared that specify autonomous driving as a requirement.

"There were a lot of great finishes in the Grand Challenge and that shifted a lot of things from intention to action."

Some of that interest is from the military. After all, DARPA, the Pentagon's research and development arm, had sponsored the race to foster innovation in autonomous ground vehicles; the armed forces are facing a mandate to make a third of their vehicles self-driving within the next 10 years.

But Dr. Whittaker, whose work 20 years ago using robots in the cleanup of the Three Mile Island nuclear accident was a milestone in establishing the concept of "field robotics," said the interest in autonomous vehicles is much broader than that, including environmental remediation, mining and agricultural machinery.

The necessary components for a self-driving vehicle -- computers, sensors -- are all available off the shelf, said DARPA Director Anthony Tether, but assembling them to produce a truly autonomous vehicle, rather than a remote-controlled vehicle, had proved daunting.

"The secret sauce," he said after the race last month, "is really the algorithm" -- the software approaches to analyzing sensor data and translating that information into driving instructions.

The Grand Challenge attracted thousands of people -- or "gray matter," in Dr. Tether's view -- to focus on that problem and find an algorithm that works.

"At DARPA, our approach is to look at ideas and concepts that people didn't think were possible and to show that they can be done," he added. "We take the technical excuse off the table. It's amazing the acceleration of progress when people see it is possible."

William Thomasmeyer, president of the National Center for Defense Robotics, said the Grand Challenge was a watershed event, demonstrating an ability to reliably follow GPS coordinates and to both detect and avoid obstacles.

Israel already has made the commitment to use unmanned ground vehicles to patrol the security fence being erected along its West Bank border, he noted. That was one of the topics that Israeli robotics firms discussed last week with area firms at a conference arranged by Mr. Thomasmeyer. Dr. Whittaker has identified a fence maintenance robot as one of his new initiatives.

Since the race, both the Stanford and Carnegie Mellon teams have talked with DARPA about the development of convoying capability -- enabling trucks to "follow the leader," and thus eliminate most, but not all, of the human drivers in military truck convoys.

Dr. Thrun, a Carnegie Mellon computer scientist before joining Stanford in 2003, said he has rejected proposals to develop armed vehicles under computer control, but is pursuing the convoy work. "In Iraq, huge numbers of people die in convoys" because of mines, he noted. Those casualties could be reduced if robots can assume more mundane, routine tasks.

Volkswagen demonstrated automated convoy driving several years ago, he said. But the system can be made more robust by, for instance, making it possible for other vehicles to insert themselves into a line of trucks without disrupting the convoy chain.

Enabling autonomous vehicles to deal with traffic of all kinds -- moving robotic vehicles from lonely desert roads to busy highways -- is one of the next big hurdles, Dr. Thrun said, and one to which he intends to devote much of his attention.

"My personal take is that the biggest impact you can have is reducing the 43,000 deaths each year of people on the highways," he said. It should be possible eventually for a vehicle to drive itself for hundreds of miles on public highways. The idea wouldn't be to eliminate human drivers, but to develop a series of driver-assistance systems that help keep the car on the road despite hazardous roads or driver inattention.

Dr. Whittaker, who assembled his 50-member team and some team sponsors last week, one month after the race, said he is in talks regarding an environmental remediation project that would require about 1,000 miles of autonomous driving. He also is pursuing work regarding an autonomous vehicle for maintaining security fences, as well as continuing work on automated mine mapping.

He also expressed his intention to move many of his activities off the Carnegie Mellon campus and onto the former LTV site in Hazelwood, where the early development of his Grand Challenge racers, H1ghlander and Sandstorm, took place over the last two years.

"Field robotics requires a field," he explained. He is still in talks with the various entitities that control the old LTV site and have previously permitted his Grand Challenge work there.

But he already has a name picked out for the facility: Robot City.