Google's Driverless Cars Are Getting Safer

Google's driverless cars are getting safer, even as they spend more time on California roads, according to research data reported this week.

Photo: Justin Sullivan/Getty Images

Photo: Justin Sullivan/Getty Images

The number of times when system failures or safety concerns prompted a test driver to take manual control of one of the vehicles running in autonomous mode dropped by 75 percent in 2016 compared with a year earlier, according to Waymo, Google's self-driving car subsidiary.

This past year, the vehicles hit one of these roadbumps 124 times, compared with 341 times in 2015. Meanwhile, the cars covered 50 percent more ground in test driving.

Each time a test driver needs to take over due to an autonomous mode failure or safety issue, it's taken as a learning experience, according to Dmitri Dolgov, head of Waymo's self-driving car technology.

"We’ve been able to make dramatic improvements to our technology because we use each of these disengages to teach and refine our car," Dolgov wrote in a blog post. "For each event we can create hundreds — and sometimes thousands — of related scenarios in simulation, varying the parameters such as the position and speed of other road users in the area."