Self-driving Uber that killed pedestrian reportedly didn’t realize she was human
The 49-year-old woman struck and killed by a self-driving Uber vehicle in Arizona earlier this year was detected by the autonomous vehicle’s sensors, but the car didn’t avoid her due to an improper software setting, according to a Monday report from the Information.
Elaine Herzberg was hit on a March night while walking a bicycle across a main thoroughfare. The Uber vehicle with an operator in the front seat was in autonomous mode at the time of the crash. “Two people briefed about the matter” told the Information that the software in the car detected the woman, but there was a problem with how the software decides how to react to objects.
The Information says the software is able to decide if an object is a “false positive,” like debris or trash that it doesn’t need to react to. It seems the software detected a false positive instead of a pedestrian. It mistakenly ignored Herzberg and didn’t react appropriately or fast enough.
Uber didn’t comment on the report about the software problem, but told Mashable they are reviewing the self-driving program and cooperating with a National Transportation Safety Bureau investigation. The company also announced Monday that it hired former NTSB chair Christopher Hart to oversee safety.
“Our review is looking at everything from the safety of our system to our training processes for vehicle operators, and we hope to have more to say soon,” an Uber spokesperson said in an email.
The report is the first tidbit about the crash since police released terrifying footage of the operator behind the wheel the moment before impact. Speculation of what contributed to the fatal crash has ranged from fewer sensors to cutting corners with training.
Uber’s self-driving program remains suspended at all its testing sites.