Driverless cars may prevent far fewer crashes than previously thought

Driverless cars have been touted as a possible solution to road traffic accidents, with research previously suggesting that up to 90 per cent of accidents could be avoided.

A new study finds that, in reality, autonomous vehicles may only reduce accidents by a third, unless the AI driving style is optimised for safety rather than speed or convenience.

Currently, mistakes made by the driver play at least some role in 9 out of 10 crashes, according to a national survey of police-reported crashes across the US.

If autonomous vehicles are designed to drive too much like people, their safety benefits will not be as pronounced, the Insurance Institute for Highway Safety (IIHS) said.

“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,” said Jessica Cicchino, IIHS vice-president for research and a co-author of the study.

“Building self-driving cars that drive as well as people do is a big challenge in itself,” said IIHS researcher 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.”

To estimate how many crashes might continue to occur if self-driving cars are designed to make the same decisions about risk that humans do, the IIHS examined more than 5,000 police-reported crashes from the National Motor Vehicle Crash Causation Survey.

This sample is representative of crashes across the US in which at least one vehicle was towed away and emergency medical services were called to the scene.

They reviewed the case files and separated the driver-related factors into different categories alongside unavoidable crashes, such as those caused by a vehicle failure like a blowout or broken axle.

They found that one-third of all crashes were the exclusive result of sensing and perception errors or driver incapacitation.

However, most crashes were due to more complex errors, such as making wrong assumptions about other road users' actions, driving too fast or too slow for road conditions, or making incorrect evasive manoeuvres. Many crashes resulted from multiple mistakes.

Self-driving vehicles will need to not only obey traffic laws, but also to adapt to road conditions and implement driving strategies that account for uncertainty about what other road users will do, such as driving more slowly than a human driver would in areas with high pedestrian traffic or in low-visibility conditions.

“Our analysis shows that it will be crucial for designers to prioritise safety over rider preferences if autonomous vehicles are to live up to their promise to be safer than human drivers,” Mueller said.

Earlier this year, another team proved that driverless systems could be tricked into emergency braking by projecting ‘phantom’ images onto the road that fool the autopilot into thinking a pedestrian is standing in the vehicle’s path.