Uber’s self-driving car that killed a pedestrian in March 2018 had severe software defects, including the inability to detect jaywalkers, according to the National Transportation Safety Board (NTSB).
The US safety agency stated that Uber’s software failed to detect the 49-year-old victim, Elaine Herzberg, as a pedestrian crossing the street. It didn’t analyze that it could potentially crash with her until 1.2 seconds before effect, at which point it was too late to brake.
Furthermore, the NTSB stated Uber’s system design “did not include a consideration for jaywalking pedestrians.” On top of that, the vehicle initiated a one-second braking delay so that the vehicle could analyze an alternative path or let the safety driver take control.
Uber has since removed that function in a software update.
Uber’s autonomous test vehicles may have failed to detect roadway hazards in at least two other incidents. In one of the cases, a vehicle struck a bicycle lane post that had bent into a roadway. In the second case, a safety driver was forced to take over to avoid an oncoming vehicle and ended up to strike a parked car. In the seven months before the fatal crash, Uber vehicles were involved in 37 accidents, consisting of 33 in which other vehicles struck the Uber test cars.
When Uber restarted testing in December 2018, it used majorly revised software. According to the NTSB, Uber did a simulation of the recent system utilizing sensor data from the fatal Arizona accident. It found that it would have detected the pedestrian 289 feet before impact and had four seconds to brake prior impact at a speed of 43.2 mph. The average stopping distance for a person is about 130 feet at that speed, including reaction time, so it seems possible that the vehicle would have been capable to stop in that distance.
The NTSB is going to meet on November 19 to figure out the cause of the accident that took place in Tempe, Arizona in March of 2018. Prosecutors have already absolved Uber of criminal liability, but are still weighing criminal charges against the driver.