In an airplane hanger on MIT’s campus in Cambridge last week, a team of engineering students and researchers put the finishing touches on Talos, a Land Rover that drives itself. Talos is MIT’s entry in the Defense Advanced Research Project Agency’s (DARPA) robotic car race, which will take place on November 3, in Victorville, CA.
Known as the Urban Challenge, the race will test the ability of robotic cars from 35 different teams to obey traffic laws and drive safely in a city-like environment without human assistance. The vehicles will need to find their way to a preprogrammed destination while paying attention to lane markers, other cars, and unexpected obstacles, such as potholes in the road. (See video.)
The Urban Challenge is a follow-up to DARPA’s Grand Challenge race, held in 2004 and 2005, in which cars navigated an empty desert road. The new, more complex racing environment reflects the rapid progress being made in robotic cars: while none of the teams finished the first Grand Challenge race, 5 out of 23 cars finished the second one. Stanford University’s team, which won the latter race, will enter the Urban Challenge with Junior, an upgraded version of its winning car. (See “Stanford’s New Driverless Car.”)
In order to “see” its environment, MIT’s Talos is equipped with numerous laser range finders, radar units, Global Positioning Systems, and video cameras, explains Emilio Frazzoli, a professor of aeronautics and astronautics and one of the team leaders. The researchers developed novel software–which runs on 10 quad-core computers in the Land Rover’s trunk–to make sense of the incoming data and to calculate the car’s next move. The 40 processors produce so much heat that the team added an air-conditioning unit to the roof of the car. (See slide show.)
See the technology that operates MIT's Land Rover.
Watch the Land Rover navigate a sample course.
Many of the robotic cars at the Urban Challenge will be outfitted with similar collections of off-the-shelf sensors, so it’s nuances in each car’s software that will likely distinguish winners from losers. MIT’s software consists of algorithms that work with the sensors to build a picture of the environment, and algorithms that determine what the car should do with that picture, explains Frazzoli. Every second, the algorithms use data from the sensors to generate more than a thousand possible paths that the car could take. Talos then drives along the path with the highest probability of producing the most direct and safest route for a given situation.
For the MIT team, which started developing Talos about a year ago, the challenge is to make sure that the car is reliable in as many different locations as possible. “We’re testing almost every day,” says Frazzoli. When the car arrives in Victorville, the team will continue to test for about a month before the preliminary trials begin. “It’s not too hard to build a robotic car,” Frazzoli says. “But it is hard to build one that’s robust and safe in many different environments.”
DeepMind’s cofounder: Generative AI is just a phase. What’s next is interactive AI.
“This is a profound moment in the history of technology,” says Mustafa Suleyman.
What to know about this autumn’s covid vaccines
New variants will pose a challenge, but early signs suggest the shots will still boost antibody responses.
Human-plus-AI solutions mitigate security threats
With the right human oversight, emerging technologies like artificial intelligence can help keep business and customer data secure
Next slide, please: A brief history of the corporate presentation
From million-dollar slide shows to Steve Jobs’s introduction of the iPhone, a bit of show business never hurt plain old business.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.