Last week at the Center for Automotive Research at Stanford (CARS), we looked at a car designed to study vehicle dynamics. Today, we get a look at how engineers and computer scientists are building cars that don’t need drivers—and are better for it.
Mike Sokolsky is a research engineer in the artificial intelligence lab at Stanford. He works as a systems integrator for “Junior,” Stanford’s car that tests computer vision, decision-making, and probabilistic planning. “If you think about a GPS device, right now you can plug in a place and it tells you how to get there,” says Sokolsky. “The extension that we want to get to is, you plug in a place, and it takes you there. That’s the interface that we want, ultimately, to be able to interact with our cars: sit back, relax, and it takes care of all the driving for you.”
What they’re working on now at CARS is taking the autonomous car from the test environment to real roads, with real problems and lots of unknowns, like construction. “You can’t rely on the roads being the same every time you come back. You have to be able to adapt to all these changing situations,” says Sokolsky.
While the interior of Junior looks pretty much like a normal car, the trunk of the car looks like a miniature data center. But computers in back are nothing special, says Sokolsky: “We’ve put a lot of effort into making sure that the software we write is scalable, and works sort of incrementally, so you don’t need to worry about massive data. There’s already a huge amount of data that comes in from all the cameras, sensors, and the laser on the roof, and to integrate all this is a big effort in itself. So we try to keep things as simple as possible and solve problems with minimal effort.”
Sokolsky estimates that many of the features that Junior employs, like cameras monitoring lanes and blind spot detection, will start becoming standard over the next ten years. But he also thinks that people will have to adapt, at least as much as the cars. “In some ways I think the technology is going to come along much faster than both the legal issues and societal acceptance,” says Sokolsky. “Because you’re going to have to convince people to give up driving their car everywhere, and some people are going to be extremely reluctant to do that. In this field, we sort of lose sight of this, but we talk to people, and they say, ‘That sounds terrifying.’ There’s a lot to overcome in terms of that.”
Stanford Automotive: http://me.stanford.edu/groups/design/automotive/
Video of Junior doing an amazing parking job: http://singularityhub.com/2010/05/12/stanfords-robot-car-slides-into-parking-spot-like-a-badass-video/