Google recently revealed that its 23 self-driving cars in California have been in 11 minor accidents since 2009, when the company began putting its cars on the road. All of the accidents involved “light damage, no injuries. To put that in perspective: the national average for “property-damage-only crashes” (that is, fender benders) is 0.3 per 100,000 miles driven.
Google’s rate is double the national average at 0.6, with 11 accidents over 1.7 million miles driven. But Google’s spokespeople are quick to note that many minor accidents go unreported, so the national average could be underestimated. This is all according to a blog post written by Chris Urmson, the director of Google’s self-driving car program.
The ultimate goal of self-driving cars is to make roads safer and lower the number of deaths from crashes. So the high accidents rate is somewhat disconcerting. But it’s also worth keeping in mind that Google has been prioritizing its efforts on the kinds of accidents that kill people, not the kind that scuffs up your car.
The fact that self-driving cars are even possible, given the myriad inputs they need to react to at any moment–the moving patterns of other cars, bicyclists darting around, stop lights, street signs–is impressive. Making that all understandable by a machine is a remarkable feat of technology. But clearly, self-driving cars still have a ways to go before they are ready for prime time.