Story at a glance
- New research finds that self-driving cars may only prevent one-third of accidents caused with human drivers.
- Successful autonomous vehicles would need to be programmed to prioritize safety over rider conveniences, experts say./li>
Autonomous vehicle technology, still in its infancy, is marketed based on the premise that its machinery can remove the human error responsible for many driving accidents and give way to safer automobile travel.
A new study calls this notion into question.
Commissioned by the Insurance Institute for Highway Safety (IIHS), researchers found that self-driving vehicles may not mitigate accidents along highway roads. The results of the study suggest that autonomous vehicles may only prevent about a third — 34 percent — of car crashes.
BREAKING NEWS ABOUT THE GEORGE FLOYD CASE
“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,” Jessica Cicchino, IIHS vice president for research and a co-author of the study, wrote.
Researchers examined more than 5,000 police reports detailing the causes of crashes sourced from the National Motor Vehicle Crash Causation Survey to estimate how many crashes would occur if self-driving cars make similar decisions regarding risk as human drivers. This sample is considered representative of all crashes across the U.S. and included at least one vehicle that had to be towed from the scene and emergency medical services were called.
Scientists then constructed a model where all vehicles on the road were self-driving and assumed the crashes they could prevent were those resulting from human perception errors, like merging and breaking on time.
The crashes due to these sensing and perception errors accounted for approximately 24 percent of all crashes, while driver incapacitation — such as driving under the influence — accounted for another 10 percent. Self-driving vehicles ideally would be equipped with sensors and cameras to monitor roads and identify potential hazards in a way human drivers cannot.
In theory, autonomous cars could prevent this 34 percent of accidents, assuming they function perfectly at all times.
The remaining two-thirds, however, are likely to still occur, since they are largely caused by planning and deciding errors, such as speeding and illegal maneuvers, which tend to be results of driver preferences and choices. Some portion of the crashes are also due to mechanical failures like tire blowouts and broken axles.
In order for self-driving cars to avoid more of the remaining two-thirds of crashes they are intended to prevent, their technology will need to be programmed to drive with safety in mind rather than speed or convenience. For example, that may mean driving slower than a human driver normally would in an area with high pedestrian traffic or during conditions of lower visibility.
“Our analysis shows that 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,” says IIHS research scientist Alexandra Mueller, the lead author of the study. “But they’d actually need to be better than that to deliver on the promises we’ve all heard.”
This comes as several companies, most notably ridesharing giant Uber, have been testing and cautiously deploying autonomous vehicles as a way to reduce costs and prove profitability.
In 2018, however, efforts came under scrutiny after an accident with a self-driving Uber resulted in the death of 49-year-old Elaine Herzberg in Tempe, Ariz.
Recently, Uber’s self-driving car department, the Advanced Technologies Group (ATG), has plans to continue testing the technology and launch in select cities that have hospitable conditions for safely deploying autonomous vehicles.
READ MORE ABOUT THE GEORGE FLOYD PROTESTS