Researchers Are Teaching Self-Driving Car Systems To Recognize Road Rage

Self-driving cars need to learn how to recognize bad drivers. Just not honk at them, like you do.

byAaron Cole|
Self-Driving Tech photo

The road to self-driving cars will almost definitely include someone trying to cut one off. That’s not a metaphor for anything; it’s literally going to happen. That’s why researchers from the University of Warwick in England are working on training self-driving cars to recognize road rage and then steer the hell away from it.

Researchers from the university published a paper this week outlining the typical responses they say angry people have behind the wheel and help future autonomous vehicles respond appropriately. According to researchers, angry people drive faster, make more mistakes, and endanger others. In other news: water is wet, and orange juice is made from oranges.

“This research is significant because, as the era of autonomous vehicles approaches, road traffic will be a mix of both autonomous and non-autonomous vehicles, driven by people that may be engaged in aggressive driving,” lead author of the study Zhizhuo Su said in a statement. “This is the first study to characterise aggressive driving behaviour quantitatively in a systematic way, which may help the autonomous vehicles identify potential aggressive driving in the surrounding environment."

Jokes aside, quantitatively measuring angry drivers’ performances will be critical to self-driving cars sharing the same roadways with people-piloted vehicles. Understanding erratic behavior, including angry or impaired drivers, will sadly be a part of the calculus required for self-driving cars to co-exist with people. Researchers say that angry drivers reportedly drove roughly 3 mph faster than the traffic around them and made about 2.5 times more errors behind the wheel than calm drivers. Researchers didn’t shake up drivers and let them loose on roads; roughly 1,800 respondents were asked to recall how they drove while angry and report what they did. That's to say: the data here could be a little sandbagged. 

It's useful nonetheless to create parameters on how self-driving systems could react to future volatile drivers, which they must do. According to the World Health Organization, roughly 1.3 million people died on the roads in 2019, and human errors caused 90-95% of those fatal crashes. Getting a better picture of what drivers do when they’re agitated can help save lives. Here’s my help: Train cameras to recognize middle fingers, then head to the nearest exit. 

Got a tip? Send it in to