Valentine's Day 2016 was important to many folks. Couples pledged their undying love, companies like Hallmark and Godiva lolled on piles of holiday dough, and in Mountain View, California, one of Google's autonomous cars hit a bus.
That third item might not seem like a big deal: Google's self-driving vehicles have been involved in accidents before. However, all of the 17 fender-benders previously reported by the tech firm have been attributed to human drivers--either drivers of other vehicles, or Google drivers who'd taken control of a self-driving vehicle for some reason.
On February 14, though, Google says that its autonomous Lexus RX 450h was partially to blame for the accident. That's a first.
According to Google, the trouble began when its car maneuvered around some sandbags in a wide lane. As the vehicle moved left, back to the center of the lane, the car's software and its driver believed that a bus traveling in the same direction would slow so that Google's car could merge into traffic. That didn't happen.
Thankfully, both vehicles were traveling very slowly: the bus was moving at 15 mph, and Google's car was traveling at 2 mph. No one in the car or on the bus was injured, though the Lexus sustained some damage to a sensor on the driver's side, as well as the left fender and left front wheel.
Google issued a statement that reads, in part, "we clearly bear some responsibility, because if our car hadn’t moved, there wouldn’t have been a collision."
Both the Santa Clara Valley Transportation Authority and the California Department of Motor Vehicles are investigating the accident, and authorities are also looking into the question of liability. That should be interesting, especially since the National Highway Traffic Safety Administration recently determined that for regulatory purposes, autonomous vehicle software is a "driver". (Though Google may just pay expenses associated with the accident out of pocket, rather than relying on insurance to do the job.)
WHAT DOES THIS MEAN FOR AUTONOMOUS CARS?
Will Google's fender-bender slow the growing momentum we've seen in the field of autonomous cars? Probably not.
For starters, this one accident doesn't change the fact that Google's autonomous cars are very, very safe. Since 2009, the company's fleet of self-driving vehicles has logged over two million miles, and at last count, they'd been involved in fewer than 20 accidents--less than one for every 100,000 miles traveled. Few human drivers could perform that well.
If anything, Google's cars have been criticized for being too cautious. If the vehicles in this particular case were traveling at higher speeds, the accident might not have happened, because Google's car may have been more hesitant to merge.
That said, it's clear that fully autonomous vehicles aren't ready for primetime. (That's why Tesla's autopilot feature is only available under specific conditions.) The upside of accidents like this is that Google engineers now have more real-world data to apply to self-driving software, helping them perfect the code for its eventual roll-out to the public.