Google's fleet of more than 20 self-driving cars have been involved in 11 accidents, the search giant has revealed.
The data, released by the project's director Chris Urmson in a blog post on Backchannel, describing them as "minor" incidents with "light damage" and "no injuries".
He also noted that "not once was the self-driving car the cause of the accident" in any of the collisions, which occurred over 1.7 million miles driven by the autonomous vehicles - a million when no-one but the car was in control.
The majority of these collisions were caused by other drivers hitting the cars from behind, and Urmson highlighted intersections as "scary places" where many serious accidents occur.
Google claims that its safety record is a result of advanced sensors, which can "keep track of other vehicles, cyclists, and pedestrians out to a distance of nearly two football fields".
The cars have also been programmed to account for human error in other road users. They will pause briefly after a traffic light turns green as "that's often when someone will barrel impatiently or distractedly through the intersection".
Google has been test-driving its fleet rigorously: Urmson claims that they're now averaging around 10,000 self-driven miles a week.
Compared with the fact that that's almost as much as the average US motorist drives in a year, a record of just 11 crashes seems pretty remarkable.
Following instructions from the USA's Department of Motor Vehicles, Google was last year forced to add brake pedals and a steering wheel to its cars in order to make them safe for driving on California's roads.
Guest editor's view
TechUK CEO Julian David says: A total 1.7 million miles and only 11 accidents - that certainly puts me to shame! And the fact that none of them was the fault of the digital car does show that technology can solve the challenges of modern living. It also provides some new justification for the traditional cry from IT professionals facing system problems - it's the users who are to blame!'
-
RSAC Conference 2025: The front line of cyber innovationITPro Podcast Ransomware, quantum computing, and an unsurprising focus on AI were highlights of this year's event
-
Anthropic CEO Dario Amodei thinks we're burying our heads in the sand on AI job lossesNews With AI set to hit entry-level jobs especially, some industry execs say clear warning signs are being ignored
-
Why cutting-edge innovation is killing the planetIn-depth AI and robots will do our work, we’ll get paid in cryptocurrency, and cars will drive themselves – but each of these technologies is a massive energy hog
-
Meet the CAVForth project: The world's first autonomous busCase Study Edinburgh's AB1 route, in Scotland, lets the public catch a ride on an self-driving bus over the Forth Road Bridge. But is this really the future of public transit?
-
UC Irvine selects Velodyne Lidar’s traffic-monitoring solutionNews HIMaC² will leverage Velodyne’s Intelligent Infrastructure Solution to improve road traffic safety and efficiency
-
Sonatus secures $35 billion to spur software-defined vehiclesNews The funding will be used to turn cars into data centers on wheels
-
Oculii raises $55 million for its AI-powered radar softwareNews The VAI platform by Oculii improves the angular resolution of any radar by up to 100 times
-
UK gov to allow self-driving cars on motorways this yearNews Automated Lane Keeping Systems to be legally defined as 'self-driving' as a cautious first step to allowing fully autonomous cars on UK roads
-
ZF augments vehicle intelligence with next-gen AI supercomputerNews New ZF ProAI packs a punch with up to 66% more computing power than its predecessor
-
Why driverless cars don't belong on public roads (yet)Opinion Autonomous vehicles still can't account for human error
