March 17, 2017

How Can We Use Data to Make Self-Driving Cars Safer?

yns-plt-153240As the data surrounding Automated Vehicles grows, safety improves but privacy concerns also arise.

The majority of human-operated vehicle accidents occur due to human errors like drunk driving or road rage. Presumably, AVs will lead to a safer driving environment by reducing, and one day, removing, the human factor from the road.

Big data is already being collected from cars with onboard systems. Only recently, MobilEye, an on-board system provider to IBM and Mercedes was acquired by Intel, Vasant Dhar, a professor and director of graduate studies at the CDS, recounts.

These onboard systems are well-rounded, constantly recording the vehicle’s speed, the weather and light, measuring the impact in an accident, and even driver awareness and attention. The human-driver aspect would be recorded by the onboard system by way of the sensors.

Say a vehicle gets too close to another or approaches the edge of the road. Those are signals that a driver is distracted. AVs have yet to learn how to behave like humans, though, before making their debut on the streets seamless. People can read other people’s driving behaviors and react accordingly, and AVs will need to understand that intuition as well.

The corresponding data collected from onboard systems continually improves at ascertaining fault in accidents. This will, in turn, modify insurance policies that currently leave liability decisions debatable. Eventually, if the onboard systems regularly determine fault accurately, it could automate the liability process.

By establishing what or who’s to blame in an accident, the onboard systems also shed light on how future accidents can be avoided. “The availability of big data obtained from onboard vehicular systems and roadway sensors changes the calculus of liability,” as Dhar sums it up in his Computer article, “Equity, Safety, and Privacy in the Autonomous Vehicle Era.”

But privacy is still a major concern when data of this magnitude is involved. A driver’s location, which is always recorded, is one example of an invasion of privacy. In general, if an insurer has access to the continually recorded data on vehicles, the driver is no longer in control of the sale of their information. Drivers going over the speed limit or rolling a stop sign would be another, less invasive, consequence of the data collection from onboard systems.

If the data is legally regulated and privacy concerns are managed, there is much to gain from the insurance policies, accident determination and accident prevention that will improve with the data from onboard systems.