On Tuesday PBS aired an episode of Nova called "The Great Robot Race." It centered on the DARPA Grand Challenge, a Pentagon-sponsored race of robotic, driverless vehicles held in the Nevada desert in March 2004.
At the center of the milestone competition are two autonomous robotic vehicles: H1ghlander, a modified Hummer created by a team at Carnegie Mellon University team headed by Red Whittaker; and Stanley (left), a (far cuter) modified Volkswagen Touareg built by Stanford University's Sebastian Thrun (also left) and his students. (Thrun is coauthor or coeditor of five, count 'em, five MIT Press books.)
The program is fascinating, and not only because of the built-in competitive atmosphere surrounding the race. (Besides the fame and glory, the wining team would take home a $2 million check.) The two teams took decidedly different strategies for navigating the grueling 130-mile course. The Carnegie Mellon team used meitculous topographic mapping of the race course, to plan in advance for what the vehicle would encounter. H1ghlander was also a highly hardware-dependent vehicle. Stanley, by contrast, relied largely on its adaptive vision software, which effectively allowed it to learn about the course as it went on.
We won't tell you who won, though an upset might have been in the cards. PBS has a well-stocked website devoted to the show here, where you can watch the episode and find out a lot about the robots. John Lithgow narrates -- could you ask for more?
P.S. - In case the answer to that last question is yes -- i.e. you are a confirmed autonomous-robot nerd -- you can check out a long article about Thrun and Stanley in the January 2006 issue of Wired.