The National Transportation Safety Board (NTSB) called on federal regulators to create a review process before allowing automated test vehicles to operate on public roads. The NTSB made the recommendation based on its investigation of a fatal 2018 collision between an Uber automated test vehicle and a pedestrian.

The NTSB said an Uber Technologies Inc. division’s “inadequate safety culture” contributed to the March 18, 2018, nighttime fatal collision between an Uber automated test vehicle and a pedestrian. The vehicle operator was uninjured in the crash, the pedestrian died.

Uber’s Advanced Technologies Group had modified the vehicle, a 2017 Volvo XC90, with a developmental automated driving system. The vehicle’s factory-installed forward collision warning and automatic emergency braking systems were deactivated during operation of the automated system. An Uber ATG operator was in the driver’s seat, but the automated system was controlling the vehicle when it hit the pedestrian at 39 mph.

Inadequate safety culture

The NTSB determined that the cause of the collision was the failure of the Uber ATG operator to closely monitor the road and the operation of the automated driving system. It said the operator was visually distracted throughout the trip by a personal cell phone.

Contributing to the crash was Uber ATG’s "inadequate safety risk assessment procedures, ineffective oversight of the vehicle operators and a lack of adequate mechanisms for addressing operators’ automation complacency." It said the lapses all were "consequences of the division’s inadequate safety culture."

Investigators examine the crash vehicle. Source: NTSBThe NTSB said that the pedestrian’s impairment at the time of the crash, coupled with crossing outside a crosswalk, contributed to the crash. Toxicological tests on the pedestrian’s blood were positive for drugs, a condition that the NTSB said can impair perception and judgment. The agency also faulted the Arizona Department of Transportation for its "insufficient oversight" of automated vehicle testing.

Accident details

The Uber ATG was completing the second loop on an established test route that included part of northbound N. Mill Avenue in Tempe, Arizona. The vehicle had been operating about 19 minutes in autonomous mode — controlled by the automated driving system (ADS) — when it approached the collision site in the right lane at a speed of 45 mph, as recorded by the ADS.

About that time, the pedestrian began walking across the road where there was no crosswalk, pushing a bicycle by her side.

The ADS detected the pedestrian 5.6 seconds before impact. Although the ADS continued to track the pedestrian until the crash, it never accurately classified her as a pedestrian or predicted her path. By the time the ADS determined that a collision was imminent, the situation exceeded the response specifications of the ADS braking system, the NTSB said. The system design precluded activation of emergency braking for collision mitigation, relying instead on the operator’s intervention to avoid a collision or mitigate an impact.

View of the self-driving system data playback at about 1.3 seconds before impact, when the system determined an emergency braking maneuver would be needed to mitigate a collision. Yellow bands are shown in meters ahead. Orange lines show the center of mapped travel lanes. The purple shaded area shows the path the vehicle traveled, with the green line showing the center of that path. Source: NTSBVideo from the Uber ATG's inward-facing camera showed that the operator was glancing away from the road for an extended period while the vehicle was approaching the pedestrian, the NTSB said. Specifically, she was looking toward the bottom of the center console, where she had placed her cell phone at the start of the trip.

The operator redirected her gaze to the road ahead about 1 second before impact. ADS data show that the operator began steering toward the left 0.02 seconds before hitting the pedestrian at a speed of 39 mph.

Among the investigation’s findings:

• The Uber ATG automated driving system detected the pedestrian 5.6 seconds before impact. Although the system continued to track the pedestrian until the crash, it never accurately identified the object crossing the road as a pedestrian or predicted its path.
• Had the vehicle operator been attentive, the operator would likely have had enough time to detect and react to the crossing pedestrian to avoid the crash or mitigate the impact.
• While Uber ATG managers had the ability to retroactively monitor the behavior of vehicle operators, they rarely did so. The company’s ineffective oversight was exacerbated by its decision to remove a second operator from the vehicle during testing of the automated driving system.
• Uber ATG made several changes to address the deficiencies identified, including implementation of a safety management system.

The NTSB recommended that the National Highway Traffic Safety Administration require entities wishing to test a developmental automated driving system on public roads to submit safety self-assessment plans before being allowed to begin or continue testing and that NHTSA should review the plans to ensure they include appropriate safeguards.