Waymo blames collision with motorcycle on human error, not AI

Waymo is pointing the robotic finger at humans.

On Oct. 19 in Mountain View, a Waymo self-driving test car hit a motorcylclist, who ended up in the hospital. Now, Waymo has published a blog post with more information about the accident. And it turns out it all came down to that intractable problem: human error. 

According to Waymo, the test driver of the self-driving car took control of the vehicle moments before the accident. Another car to the left “abruptly” entered the Waymo car’s lane, so the Waymo test driver took over. The driver moved to the right to avoid a collision. But unfortunately, in moving to the right lane, the driver didn’t notice the motorcycle that had just moved into that lane. So, crash.

Waymo is actually viewing the accident as an example of something that self-driving technology could have prevented. Waymo says its AI sees in 360 degrees. In simulations of the situation, the AI would have been aware of the motorcyclist. So the car would have slowed down, rather than move to right.

“People are often called upon to make split-second decisions with insufficient context,” the blog reads. “Incidents like this are what motivate all of us at Waymo to work diligently and safely to bring our technology to roads, because this is the type of situation self-driving vehicles can prevent.”

A Waymo car was in another crash in May. But that car was also in manual mode, and another driver who swerved into oncoming traffic caused the crash. Other accidents involving Tesla and Uber self-driving technology have resulted in fatalities. 

“Testing on public roads is vital to the safe development of self-driving technology,” Waymo wrote. “We’re sorry that a member of the community was injured in a collision with one of our cars.”