Uber's Self-Driving Cars Couldn’t Detect People Outside of Crosswalks

"The software inside the Uber self-driving SUV that killed an Arizona woman last year was not designed to detect pedestrians outside of a crosswalk, according to new documents released as part of a federal investigation into the incident," report Aarian Marshall and Alex Davies. But the documents, posted by the National Transportation Safety Board, point to the bigger problem of self-driving technology not being able to understand and anticipate how humans actually behave.

In addition to safety training issues, the documents suggest a series of software-related detection, braking, and vehicle speed problems that contributed to the crash. "When the car first detected [Elaine Herzberg’s] presence, 5.6 seconds before impact, it classified her as a vehicle. Then it changed its mind to 'other,' then to vehicle again, back to 'other,' then to bicycle, then to 'other' again, and finally back to bicycle," say Marshall and Davies.

They note that road design was also a contributing factor in the crash since Herzberg was crossing with her bike "near a pathway that appeared purpose-built for walkers, but was 360 feet from the nearest crosswalk.” The NTSB will hold a meeting later this month about the incident with subsequent plans to release its full report.