Mobileye wants self-driving cars to prove they're safe
Mobileye proposes a system where self-driving cars could report collisions, with mathematical formulas to determine who, or what, is at fault.
Wayne CunninghamManaging Editor / Roadshow
Wayne Cunningham reviews cars and writes about automotive technology for CNET's Roadshow. Prior to the automotive beat, he covered spyware, Web building technologies, and computer hardware. He began covering technology and the Web in 1994 as an editor of The Net magazine.
As self-driving cars begin sharing the roads with human drivers, there will inevitably be collisions. Last year,
's experimental cars suffered a couple of notable accidents, one due to the car's programming, and one caused by a human red-light runner.
Mobileye, a camera sensor company owned by Intel, published a white paper authored by professors Amnon Shashua and Shai Shalev-Shwartz, showing how it can evaluate collisions involving
to determine who, or what, is at fault. Mobileye calls the proposed system Responsibility-Sensitive Safety (RSS), and it's designed to build societal trust in this new technology.
Self-driving cars are being developed by a wide range of companies, from automakers to tier one equipment suppliers to big technology players, such as Google and
. The technology could reduce or eliminate the over 1 million deaths caused by cars around the world each year. Many self-driving cars, manned by a human safety driver, are currently being tested on public roads, and they are expected to enter regular service as robo-taxis or private cars in 2020.
There is no current standardized system for reporting accidents involving self-driving vehicles. "Previous collisions have created a lot of confusion about AVs," said Dan Galves, chief communications officer for Mobileye. He points out that companies test self-driving cars by running them for as many miles as they can, then look back at logs to determine what happened in any incidents.
Mobileye's RSS would be a standardized system used by private companies, municipalities and government regulators. It includes mathematical models for different driving scenarios to determine if the self-driving car was at fault in a collision. For example, it defines safe following distances. If a self-driving car were to rear-end another car, the RSS system would show if it wasn't maintaining a safe distance. RSS also takes into account when another car cuts into a lane in front of a self-driving car.
The system relies on the fact that self-driving cars contain a multitude of sensors, recording video from cameras and imagery from radar and lidar. Self-driving cars also can report their exact telemetry, including speed, steering angle and braking reaction time.
That data is a lot more precise than what a human driver can say about a collision.
Mobileye's RSS is merely a proposal at this point, and will require acceptance from government entities and private sector companies to be put into effect. Last week, California's Department of Motor Vehicles announced its rules for self-driving car development, including how incidents should be reported. Last month, the US Department of Transportation published its own guidelines about the cars themselves, leaving rules of the road to states.