Merging is a nightmare. I’ve driven in California, Arizona, Texas, Colorado, Utah, New Mexico, New York, South Carolina, and several other states. It sucks everywhere and all comes down to a fundamental question: should I wait, or should I go?
Humans tend to decide quickly and intuitively, relying on social interaction training from birth. Self-driving cars, however, don’t have such training and still struggle with these social interactions in traffic.
New research at the University of Copenhagen’s Department of Computer Science demonstrates this lack. Scientists analyzed videos of self-driving cars in various traffic situations uploaded to YouTube. The results show that autonomous vehicles have difficulty deciding when to yield.
“The ability to navigate in traffic is based on much more than traffic rules. Social interactions, including body language, play a major role when we signal each other in traffic,” says Professor Barry Brown, who has studied the evolution of self-driving car behavior. “This is where the programming of self-driving cars still falls short. That is why it is difficult for them to consistently understand when to stop and when someone is stopping for them, which can be both annoying and dangerous.”
The media is full of stories about how well autonomous vehicles perform. Waymo, Cruise, and Tesla, among others, have launched self-driving cars in the U.S. and Canada. Brown and team claim that actual performance is a well-kept trade secret that few have insight into.
They used YouTube footage filmed by backseat drivers in autonomous cars. One example shows a family of four standing by the curb of a residential street. There is no crosswalk. As the driverless car approaches, it slows, causing the two adults in the family to wave their hands as a sign for the car to drive on. Instead, the car stops right next to them for 11 seconds. Then, as they begin to cross the road, the car starts moving again, causing them to jump back onto the sidewalk.
Self-driving cars in San Francisco, California, have caused traffic jams and other issues, particularly in foggy conditions, creating resistance among city residents.
Brown believes the difficulty lies in that social element. “…we take the social element for granted. We don’t think about it when we get into a car and drive – we just do it automatically. But when it comes to designing systems, you need to describe everything we take for granted and incorporate it into the design.”