It was literally the dawn of a new day on October 8, as five teams crossed the finish line in the DARPA (Defense Advanced Research Projects Agency) Grand Challenge, an event that proves autonomous ground vehicles can travel long distances over difficult terrain at militarily relevant speeds.
As the sun rose in the Mojave Desert, 23 teams lined up at the beginning of the 131.6-mile course to meet the challenge. Nearly seven hours later, Stanford University's Stanley crossed the finish line, with four other teams following suit.
Stanford took home the $2 million prize for completing the entire course in six hr, 53 min, and 58 sec. Carnegie Mellon University followed close behind with its two entries — Red Team's Sandstorm finished in 7 hr, 4 min, and 50 sec, and Red Team Too's Highlander was just 10 minutes behind at 7 hr and 14 min. In fourth place was The Gray Team's Kat-5, with a time of 7 hr, 30 min, and 16 sec. OshKosh Truck's 16-ton robot Terra-Max finished the course on Sunday with an unknown time.
Stanford attributes a great deal of its success to a variety of motion control technologies. Stanley is a reinforced Volkswagen Touareg equipped with a custom drive-by-wire system, servo-controlled steering, a sensor rack, and computing system, said team member Gabriel Hoffman, a Ph.D. student in the Aero/Astro Department at Stanford.
Like almost all of the competing teams, the vehicle is fitted with five Sick laser-measurement systems for environmental sensing. Sick's LMS sensors offer accurate distance measurement and collision control throughout a scanning field of 180°.
“At the sensing level, we were able to figure out what the drivable route was by determining the trajectory and speed to follow,” Hoffman said.
The laser system was complemented by radar and vision systems via a standard video camera. Together they determined what speed the vehicle could travel — no impediments on the course allowed travel up to 35 mph while a binary flag on the system slowed the vehicle to 25 mph, letting the lasers see what was around them.
Hoffman also said Stanford set itself apart from the competition by focusing on artificial intelligence software development. A modular system of six Pentium M-Blade computers feature synchronized clocks for time stamping. Inter Process Communication (IPC) enabled communication via TCP/IP messages over the local Ethernet.
With no teams completing the course last year, Grand Challenge organizers were thrilled that five teams succeeded this year. The teams are the first ground vehicles to travel such a far distance at relatively high speed within a specified time frame. Stanley's average speed was 19.1 mph, followed by the others between 17 and 18 mph.
DARPA's mission is to accelerate development of promising technologies and turn them over to others for development in viable applications.
“These vehicles haven't just achieved world records, they've made history,” said DARPA director Dr. Tony Tether. “We have completed our mission here, and look forward to watching these vehicles take off.”