Autonomous Urban Driving

Seth Teller
Computer Science and Artificial Intelligence Laboratory

Graduate students from across the Institute and including EECS participated in the extensive field testing to prepare the MIT autonomous vehicle for the final urban grand challenge in Fall 2008.

Graduate students from across the Institute and including EECS participated in the extensive field testing to prepare the MIT autonomous vehicle for the final urban grand challenge in Fall 2008.

Several EECS faculty and students, along with others from Mechanical Engineering, Aero/Astro, CSAIL, LIDS, Draper Laboratory, Lincoln Laboratory, and Olin College, competed in the 2007-08 “DARPA Urban Challenge,” a follow-on to the first two DARPA Grand Challenges. The goal of the Urban Challenge was the development of a safe, driverless (i.e., fully autonomous) fullsize passenger vehicle able to drive safely among other vehicles in an urban-like road network. MIT was one of eighty-nine teams to enter the challenge in May of 2006, and was one of only eleven teams judged by DARPA as qualified to compete in the final round of competition in November, 2007. Of those eleven teams, five were removed from the course during competition due to unsafe behavior. Team MIT completed the competition safely, one of only six teams to do so. The top three finishers (CMU, Stanford, and Virginia Tech) were awarded cash prizes; MIT finished fourth.

We were enormously excited to develop an autonomous vehicle, not simply because of the competition and the sheer engineering challenge, but because of the major societal benefits that will come with widespread deployment of robotic cars. Foremost is safety: self-driving cars will eliminate many of the more than one million roadway deaths, and tens of millions of injuries, suffered by human drivers annually world-wide.

Second is energy: self-driving cars can change speed more smoothly and react more quickly than human drivers (requiring less wasteful braking), achieving higher fuel efficiency. Coordination with other vehicles and the roadway infrastructure will yield further efficiency gains, through (for example) signal lights that turn green for approaching vehicles, self-parking cars that find the nearest open parking space, and car-pooling and suburban transit systems that become so convenient (with automatic passenger drop-off and pickup, and self-parking) that they are widely adopted. Third is productivity: with no need to concentrate on driving, commuters will be able to work, read, or relax while en route.

MIT’s self-driving Land Rover takes to the road as it is prepared for the final competition in the 2007 DARPA Urban Challenge. (photo: Jason Dorfman/CSAIL).

MIT’s self-driving Land Rover takes to the road as it is prepared for the final competition in the 2007 DARPA Urban Challenge. (photo: Jason Dorfman/CSAIL).

Though the DARPA Challenge ended in November 2007, MIT continues to push hard on the development of a variety of safe, driverless vehicles. We are pursuing several follow-on projects, including mechanisms for passenger vehicle collision avoidance and mitigation, and the development of a voice and gesture-commandable autonomous forklift capable of outdoor operation.

The team is now developing a voiceand gesture-commandable autonomous forklift capable of outdoor operation.

The team is now developing a voiceand gesture-commandable autonomous forklift capable of outdoor operation.

Leave a Reply

Replies which add to the content of this article in the EECS Newsletter are welcome. The Department reserves the right to moderate all comments. If you would like to provide any updated information please send an email to newsletter@eecs.mit.edu.