Two Robot Vehicles Set for $2 Million Desert Race

December 1, 2004

Two robotic vehicles using radar systems provided by students at Duke’s Pratt School of Engineering performed well in initial qualifying runs and appeared likely to be selected to compete in the $2 million Grand Challenge race sponsored by the Defense Advanced Research Projects Agency (DARPA) on Oct. 8.

The two sensor-laden, driverless vehicles were developed by a team led by the Robotics Institute of Carnegie Mellon University. Called the Red Team, the group is an alliance of individuals, non-profit institutions and corporations formed to navigate a rugged, 150-mile desert obstacle course starting and ending in Primm, Nev., with autonomous vehicles of any design.

Each robot uses Global Position Systems and inertial measurements to determine position and each perceives terrain with laser sensors, cameras and the radar. The robots use computers to calculate the course, set pace and react to contingencies.

The Pratt students, members of the Duke Robotics Club, turned commercial radar systems into short and long range sensors integrated into the “brains” of the robot vehicles. The students were assisted by engineers from the Boeing Co., and funded by Science Applications International Corp. (SAIC).

“The students spent about a year developing and testing the radar system, and they have done a tremendous job,” said Robert Kielb, a team adviser and senior research scientist in Pratt’s Department of Mechanical Engineering and Materials Science. “We have high hopes that the Red Team will win, aided by the Duke radar.”

Forty-three vehicles were selected by DARPA to compete in the qualifying runs that began Sept. 28 at the California Speedway in Fontana, Calif. When the qualifying runs end Oct. 5, DARPA will announce a field of 20 finalists that will compete for the $2 million prize.

On the first qualifying run, 41 vehicles competed and 11 completed the run, including both of the Red Team vehicles, H1ghlander, a modified Hummer, and Sandstorm, a Hummer-like vehicle. On H1ghlander’s second run Sept. 30, the driverless vehicle ran a 2.2 mile route in 9 minutes 22 seconds. Sandstorm ran its second qualifier in 8:59, one of the best of the run.

The first DARPA Grand Challenge was held last year, and no vehicle finished the course although the Red Team’s Sandstorm turned in the best performance. It was not equipped with radar last year.

DARPA Program Manager Ron Kurjanowicz said the initial train runs this year showed that the competing teams had made major progress during the past year. “These teams already have done the impossible,” he said.

The DARPA Grand Challenge was created in response to a Congressional and Department of Defense mandate intended to accelerate research and development in autonomous ground vehicles that could help save American lives on the battlefield.