Update, Nov. 5:
Cornell didn't win, but achieved remarkable success in the 2007 DARPA Urban Challenge. Team Cornell's self-driving Chevy Tahoe was one of only six vehicles out of 35 to successfully complete all three missions involving some 60 miles of driving on an urban course, while surviving a collision with MIT's robot car. Carnegie Mellon University, Stanford and Virginia Tech took first, second and third place, bringing home cash prizes. DARPA did not rank the remaining teams.

A Cornell team is among 11 finalists chosen by the Defense Department's Defense Advanced Research Projects Agency (DARPA) to compete in the finals of the DARPA Urban Challenge Saturday, Nov. 3, at 8 a.m. Pacific time on a closed course at the former George Air Force Base in Victorville, Calif.

The contestants' self-driving cars will attempt to complete a simulated military supply mission in an urban environment, obeying traffic laws and avoiding other robotic and human-driven cars, pedestrians and whatever else DARPA adds to the course.

The Cornell team of 15 students and four faculty members were chosen as finalists yesterday (Nov. 1) from 36 semifinalists after several days of qualifying tests.

Top prize in the competition, awarded for the fastest but still safe car, is $2 million; second prize is $1 million and third prize is $500,000. The event will be Webcast at http:// www.grandchallenge.org.

The teams will attempt to complete a complex 60-mile urban course in less than six hours. Traffic on the course will include 50 human-driven vehicles and the other robots. Speed is not the only factor in determining the winners, as vehicles must also meet the same standards required to pass the California driver's license road test.

Team Cornell has been at work for a year, converting a stock 2007 Chevy Tahoe into a drive-by-wire vehicle, equipping it with video cameras, radar and lidar sensors, and building an artificial intelligence capable of interpreting sensor input and reacting to it in the context of a preprogrammed mission. The car's vision system can identify and follow roads and read traffic signs, even including a stop sign painted on the pavement.

Many of the students have been living with the project at the abandoned Seneca Army Depot outside Romulus, N.Y., which provided a network of streets for preparation for the competition, in which the robot cars must demonstrate their ability to navigate parking lots, merge into traffic, take their correct turn at four-way stops and reroute around blocked roads.

DARPA had planned to narrow the final field to 20 teams, but more cars than expected were eliminated for safety reasons, according to DARPA director Tony Tether. Some made dangerous turns, some veered off the course, and a couple actually crashed.

Other finalists, in addition to Team Cornell, are Victor Tango (Virginia Tech), CarOLO (Technischen Universität Braunschweig), Ben Franklin Racing Team (University of Pennsylvania and Lehigh University), Stanford Racing Team, Tartan Racing (Carnegie Mellon University), MIT, Team UCF (University of Central Florida), Team Annieway (A German consortium), Intelligent Vehicle Systems (a team from Delphi, Ford and Honeywell) and Team Oshkosh (an industrial-university consortium including the Oshkosh Truck Corp.).