Adobe Stock

Autonomous vehicles (AVs) will likely prevent only about one-third of crashes if systems drive like people, according to a study from Insurance Institute for Highway Safety (IIHS).

“It’s likely that fully self-driving cars will eventually identify hazards better than people, but we found that this alone would not prevent the bulk of crashes,” says Jessica Cicchino, IIHS vice president for research and a coauthor of the study.

Safety advocates have promoted the idea that AVs are safer than human drivers because surveys of police-reported crashes show that driver error is the cause in more than 90% of accidents. However, IIHS’ analysis of crash data suggests AVs’ better optics and sensors, and their inability to experience blackouts or other medical emergencies, wouldn’t be enough to avoid two-thirds of crashes. Becoming significantly better drivers than humans would require programming that prioritizes safety ahead of speed and convenience.

“Building self-driving cars that drive as well as people do is a big challenge in itself,” says IIHS Research Scientist Alexandra Mueller, lead author of the study. “But they’d actually need to be better than that to deliver on the promises we’ve all heard.”

IIHS researchers examined more than 5,000 police-reported crashes to the National Highway Traffic Safety Administration’s (NHTSA’s) National Motor Vehicle Crash Causation Survey in which at least one vehicle was towed away, and emergency medical services were called to the scene.

Researchers categorized crashes using five common errors:

  • Sensing and perceiving – Driver distraction, impeded visibility, failing to recognize hazards
  • Predicting – Misjudging traffic gaps, incorrectly estimating other vehicles’ speeds, faulty assumptions about what another road user was going to do
  • Planning/deciding – Driving too fast/slow for road conditions, driving aggressively, leaving too little following distance
  • Execution/performance – Inadequate/incorrect evasive maneuvers, overcompensation, mistakes in controlling vehicle
  • Incapacitation – Alcohol/drug impairment, medical problems, falling asleep

Researchers also determined some crashes were unavoidable, such as those caused by a vehicle failure such as a blowout or broken axle.

To predict the impact AVs will have on safety, IIHS experts imagined a future with only self-driving vehicles. Future vehicles would prevent crashes caused by perception errors or involving an incapacitated driver.

Crashes due to only sensing and perceiving errors accounted for 24% of the total, and incapacitation accounted for 10%. Those crashes might be avoided if all vehicles were self-driving, though it would require sensors that worked perfectly and systems that never malfunctioned.

Consider the Uber test vehicle crash that killed a pedestrian in Tempe, Arizona, in March 2018. Its automated system struggled to correctly identify 49-year-old Elaine Herzberg on the side of the road. Once it did, it still was not able to predict that she would cross in front of the vehicle, and it failed to execute the correct evasive maneuver to avoid striking her.

Planning/deciding errors contributed to about 40% of crashes. When humans decide how aggressively to drive, preferences sometimes conflict with safety priorities. To eliminate most crashes, AVs will have to do better.

“It will be crucial for designers to prioritize safety over rider preferences if autonomous vehicles are to live up to their promise to be safer than human drivers,” Mueller says.

Insurance Institute for Highway Safety (IIHS)