Obama Administration Wants to Spend $4B on Self-Driving Cars

PHOTO: A self-driving car traverses a parking lot at Googles headquarters on Jan. 8, 2015 in Mountain View, Calif. PlayNoah Berger/Getty Images
WATCH Obama Administration Wants to Spend $4B on Self-Driving Cars

The Obama administration wants to kick the push for driverless cars into high gear -- proposing to spend nearly $4 billion on development over the next decade, Transportation Sec. Anthony Foxx announced today.

"We are bullish on automated vehicles," said Foxx, flanked by auto executives at the North American International Auto Show in Detroit.

Automakers and safety regulators alike hope self-driving cars could eventually eliminate the 94 percent of fatal crashes that involve some type of human error. They also believe autonomous technology could help reduce greenhouse gas emissions and loosen gridlock.

Over the next six months, the National Highway Traffic Safety Administration will come up with a prototype policy and develop best-practices for automakers, Foxx promised today. They’ll also propose new regulatory interpretations, and are encouraging automakers to request regulatory exemptions as soon as they “believe automation advances can be deployed safely.”

The DOT “wants to work with your progress, not hold it back,” Foxx said.

Though driverless technology has grown increasingly popular, autonomous vehicles aren’t perfect yet.

In a report submitted to the California DMV, Google acknowledged that its driverless cars “disengaged” 272 times between September 2014 and November 2015.

That’s the car telling the driver,“‘I don’t like something that’s going on, you drive now,’” program director Chris Urmson explained at a Transportation Research Board luncheon on Wednesday.

Another 69 times, the driver felt compelled to take the wheel even though the software hadn’t disengaged, the report said.

Google determined 13 of those 69 events were what the company called “simulated contacts” –- in other words, the simulator determined the car would have crashed into something if the driver hadn’t taken over.

By now, normal traffic patterns create few headaches (computer chip aches?) for driverless cars. It’s relatively easy to code for situations you can anticipate. What’s difficult to code for is unpredictable events – say, a woman in an electric wheelchair chasing a flock of ducks in circles mid-road. (Yes, that really happened in front of a google car, Urmson said.)

But such hiccups haven’t dissuaded automakers from major investments in driverless tech.

Just last week, General Motors announced a $500 million investment in Lyft, in part to develop a fleet of on-demand autonomous cars.