The National Transportation Safety Board’s investigation of the 2018 fatal accident in Tempe, Ariz., involving an Uber self-driving car and a pedestrian offered one finding that should give all — especially Pittsburghers — reason for pause: The vehicle wasn’t able to recognize a jaywalker.
That’s a sobering thought for pedestrians in Pittsburgh where jaywalking is ubiquitous and Uber continues testing its autonomous vehicles.
The Tempe accident led Uber to shut down its nationwide testing of self-driving vehicles for several months; testing resumed in Pittsburgh last December. Given this new information, the local testing demands even more scrutiny and oversight.
The NTSB determined that software in the Uber Technologies Inc. test vehicle spotted the pedestrian almost six seconds before the crash but didn’t engage the braking system until only a second before striking her. A primary reason was because the 49-year-old woman who was walking a bicycle across the road was not in a crosswalk and the vehicle could not recognize a jaywalker. The backup system — a vehicular operator who stands by in cases of emergency — didn’t work either.
In such a situation, Uber expected the vehicle operator to take control and override the self-driving software. The distracted operator failed to react in time and the deadly collision occurred.
The NTSB also pointed to a larger problem at play in this incident: automation complacency. In essence, the vehicle operators, who are supposed to be ready to take over if needed, become so complacent with the technology they fail to maintain an appropriate level of attention.
A software system that doesn’t recognize jaywalkers coupled with operators whose attention level could be lacking is a recipe for disaster on crowded city streets. Especially streets in Pittsburgh, where jaywalkers are as common as crooked intersections.
First Published: December 11, 2019, 10:43 a.m.