Advertisement

Google's self-driving cars master tricky three-point turns

And it only took several thousand tries per week.

REUTERS/Alexandria Sage/File Photo

With supercomputers for brains and perfect 360-degree vision, one would imagine it would be fairly simple for an autonomous vehicle to pass the time-honored driver's ed test of performing a three-point turn. But according to Google's self-driving car project October report, that is one of the many things that's easier said than done for a robot vehicle.

While a human driver can estimate the best angle and distance, the amount of data and information the autonomous vehicle has can actually be a drawback. "Our challenge is to teach our self-driving cars to choose the option that's not only the quickest, but one that feels natural to passengers," the report says. So, even though some turns would be easier done in reverse, for example, people prefer to travel moving forward, where they can see what's happening. "So we've taught our cars to mimic these human patterns, favoring wider forward arcs, rather than a series of short movements back and forth."

In order to get all these turns right, Google's vehicles are doing a lot more turns than most drivers will probably ever practice before hitting the road – about 1,000 every week. And, despite some hang ups with California's autonomous vehicle regulations, Google's vehicles have driven over 2.2 million autonomous miles in Washington State, California, Arizona and Texas to date.