SAN FRANCISCO (Reuters) - In what sounds like a science fair project on steroids, engineers at Stanford University plan to have an unmanned robot car ready to navigate urban traffic in less than a year.
The car, a 2006 Volkswagen Passat wagon dubbed Junior, is Stanford’s newest competitor in a high-stakes road race sponsored by the Defense Advanced Research Projects Agency, the U.S. Department of Defense’s research and development arm.
The Stanford car will compete in the agency’s third and most challenging derby — the DARPA Urban Challenge, in which robotic cars will drive in a mock city environment. Cars must merge, navigate traffic, traverse busy intersections, avoid obstacles and master the most delicate of skills — determining who has the right of way.
“These cars are driven by artificial intelligence,” said Sebastian Thrun, a computer science and electrical engineering professor at Stanford, who unveiled his plans for Junior this week at the annual meeting of the American Association for the Advancement of Science in San Francisco.
Stanford and Thrun have been down this road before. Stanford’s entry in the 2005 race, Stanley, won first place. But that race was run in the Nevada desert. “The next challenge will be to drive where we live,” said Thrun, who spoke on a panel about the future of robotics.
“This new generation of robots is making the case that they can safely navigate without any human assistance,” he said.
An array of other U.S. universities, many with corporate partners, are involved in the 2007 challenge, including Carnegie Mellon University, which finished a close second to Stanford in the 2005 race.
Thrun says the project may pave the way for a future in which self-driving cars will make transportation safer for those who, like the elderly, might rather ride than drive.
“By 2030, we should be able to deploy this technology on highways reliably,” he said.
Thrun said he expected a battlefield version of the car to be available as early as 2015.
Driving in a city environment means the robot cars must not only detect obstacles, they must make sense of them.
“To be able to understand your environment, predict what happens next and be able to react when something goes slightly wrong — that is the most challenging,” said Stanford research engineer Mike Montemerlo, who spoke at the meeting.
Junior’s steering, throttle and brakes have been modified by engineers at the Volkswagen of America Electronics Research Laboratory in Palo Alto, California.
The car sports an array of sophisticated sensors, including a range-finding laser that provides a three-dimensional, 360-degree view of its surroundings in near real-time.
Junior’s computer “brain” is about four times more powerful than Stanley’s was in 2005.
To make the car “think,” about a dozen students, faculty and researchers at Stanford worked on software to manage driving tasks like perception, mapping and planning.
The location for the November 3 race will be announced in October. For the fastest car to navigate the course, the prize is $2 million — plus bragging rights.