Skip to Content
MIT News magazine

Ladies and Gentlemen, Start Your Bots!

MIT fares well in DARPA’s famous robotic-car race.
December 18, 2007

It could’ve happened to anyone. Just as the Land Rover began passing a Chevy Tahoe that had stopped at the entrance to a traffic circle, the Tahoe started moving. The resulting collision was ruled “no fault.” But it was also “no driver”: it occurred in the finals of the Urban Challenge robotic-car race on November 3 in Victorville, CA, supplying one of the event’s biggest thrills. The automated Land Rover, known as Talos, was MIT’s first faculty-student entry in a U.S. Defense Advanced Research Projects Agency (DARPA) Challenge race. Talos took the fender bender with Cornell’s vehicle in stride and went on to a respectable fourth-place finish behind vehicles designed by veteran teams from Carnegie Mellon, Stanford, and Virginia Tech.

MIT’s robocar posed for a photo before its journey to California for DARPA’s Urban Challenge.

The initial DARPA Grand Challenge races, held in 2004 and 2005, tested robotic cars mainly on lonely desert roads. In a testament to the advances in robotics in the past three years, DARPA staged the Urban Challenge on an abandoned military base to assess how well driverless cars could navigate through a mock city while obeying traffic laws and interacting safely with other bots and human drivers.

To enable Talos to drive on its own, the MIT team outfitted it with hundreds of thousands of dollars’ worth of equipment, bolting on laser range finders, GPS hardware, and video cameras to capture information about the vehicle’s surroundings. To crunch all that data, the researchers installed a minia­ture supercomputer with 40 cores, stowing it behind the back seats. They also added an extra generator to keep the hardware up and running and a second air conditioner to cool the computers.

During the race, each vehicle appeared to drive with its own personality: some accelerated quickly, while others were more hesitant. Talos seemed confident at some points and tentative, as though pausing to “think,” at others. Team leader John ­Leonard, a professor of mechanical and ocean engineering, says that Talos hesitated when GPS data conflicted with sensor observations, which its designers thought should be able to overrule the navigation system. While other cars relied heavily on GPS receivers, the MIT team wanted Talos to be able to use other means to find its way along the course. “In a military setting, GPS can be jammed,” says Leonard. “Our ­vehicle might have looked cautious, but what’s going on inside is that our vehicle was looking at the road and traffic with other sensors.”

Leonard says the race was a learning experience for him and his team. “I’m tremendously proud of our team for creating such an impressive machine that qualified for this race,” he says. DARPA invited 35 teams out of 89 entrants to compete in the semifinals; of the 11 vehicles qualifying for the finals, MIT’s was one of just six to complete the challenging 55-mile course. The team plans to eventually share a large portion of its software online, along with the gigabytes of data collected during the race, so anyone can use it to develop future robotic cars.

Keep Reading

Most Popular

Large language models can do jaw-dropping things. But nobody knows exactly why.

And that's a problem. Figuring it out is one of the biggest scientific puzzles of our time and a crucial step towards controlling more powerful future models.

OpenAI teases an amazing new generative video model called Sora

The firm is sharing Sora with a small group of safety testers but the rest of us will have to wait to learn more.

Google’s Gemini is now in everything. Here’s how you can try it out.

Gmail, Docs, and more will now come with Gemini baked in. But Europeans will have to wait before they can download the app.

This baby with a head camera helped teach an AI how kids learn language

A neural network trained on the experiences of a single young child managed to learn one of the core components of language: how to match words to the objects they represent.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at customer-service@technologyreview.com with a list of newsletters you’d like to receive.