Intel powers Stanford robocar entry

Modified Volkswagen Passat to enter Darpa Urban Challenge in November

Tom Sanders at Research At Intel Day in Santa Clara, CA, vnunet.com 21 Jun 2007

Intel is betting that two of its quad-core processors will give a robot vehicle the brains to navigate a real urban area, complete with oncoming traffic and other obstacles.

The chipmaker has sponsored a team from Stanford University in the Darpa Urban Challenge.

Stanford's modified Volkswagen Passat, nicknamed 'Junior', is scheduled to take to the streets in an undisclosed location on 3 November. Intel supported the Stanford team that won the challenge in 2005.

The vehicle is loaded with laser range sensors that detect obstacles. It also combines a GPS receiver with inertial measurement tools that sense acceleration and rotation to determine position.

The Darpa Urban Challenge is the third in a series of robotics contests sponsored by the US Defence Advanced Research Projects Agency.

The first two challenges took place on a desert course and required contestants to navigate 142 miles to the finish. The first challenge in 2004 yielded no finishers, but Stanford took home the $2m first prize in October 2005.

Scheduled for 3 November, this year's Urban Challenge requires contestants to follow traffic rules and avoid other road users.

The addition of moving objects to which the car has to respond makes the challenge significantly more difficult.

"The perception of everything in your environment is what is making this hard to do," said Scott Ettinger, a researcher with Intel.

Tim Hilden, a researcher with Volkswagen, added: "Predicting the future, predicting where other objects are going to be in a few seconds, is the hardest part."

The addition of traffic rules and moving obstacles is causing a dramatic increase in the amount of data the robocars have to process.

Stanford's Junior is powered by two servers running Linux on Intel's Core-2 quad processors. The software that processes the input from the sensors and handles navigation is custom built.

Despite the added challenges, Ettinger claimed that the technology that enables the robocar is pretty much done and will soon be ready for the real world.

"Technologically we are probably not that far away. I think the social aspect is going to be the difficult part: dealing with the legal issues and getting people used to the idea. I see that as a bigger barrier than the technical issues," he said.