Tesla Motors Inc. faces a lawsuit related to its Autopilot system, after reports surfaced that a man killed in a crash in China had activated the driver-assist feature of his car.
Bart Selman, an artificial intelligence and computer science expert at Cornell University, says despite recent incidents, self-driving cars – much like commercial flights – will become ‘extraordinarily safe.’ Selman believes equipping autonomous vehicles with more sensors like Lidar, could prevent crashes down the road.
Bio: https://www.engineering.cornell.edu/research/faculty/profile.cfm?netid=bs54
Selman says:
“The video of the crash near Hong Kong last January suggests the vision system was most likely the failure. Computer vision systems are trained on large amounts of pre-labeled driver video. A rare event, such as a cleaning truck driving slowly on the side of the road, is difficult to learn correctly because there are few or no examples in the videos used to train the vision system. Adding more sensors to the car, such as Lidar, which operates like a radar for nearby object detection, is most likely the best way to avoid these types of crashes in the future.
“Overall, there is some risk that the public will become wary of self-driving technology. However, I think the safety of self-driving cars will still improve further quite rapidly, making it soon ten to one hundred times saver than human driving. We will also start to see more videos of cases where the driver-assist technology saves lives.
“In an interesting recent case, a Tesla vehicle helped drive its owner to a hospital for a medical emergency. There is an interesting analogy with the introduction of commercial flights. The airline industry worked hard to make flying extraordinarily safe. We will see a similar development in the auto industry with regards to self-driving technology.”
Cornell University has television, ISDN and dedicated Skype/Google+ Hangout studios available for media interviews.
- 30 -