Self-driving cars shown to lack social intelligence in traffic

Should I go or give way? This is one of the most basic questions in traffic, whether merging on a motorway or at the doorstep of a metro. Decisions are what humans usually make quickly and instinctively, because doing so depends on social interactions trained from the time we start walking.

On the other hand self-driving cars, which are already on the road in many parts of the world, still struggle when navigating these social interactions in traffic. This has been revealed in new research conducted at the Computer Science Department of the University of Copenhagen. The researchers analyzed videos uploaded by YouTube users of self-driving cars in various traffic conditions. The results suggest that self-driving cars have a particularly difficult time understanding when to ‘yield’ – when to give way and when to drive away.

“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 to each other in traffic. This is where the potential for self-driving cars The programming still falls short. That’s why it’s constantly difficult for them to know when to stop and when someone is stopping for them, which can be both frustrating and dangerous.” five years.

Sorry, it’s a self-driving car!

Companies such as Waymo and Cruise have introduced taxi services with self-driving cars in some parts of the United States. Tesla has rolled out its FSD model (Full Self-Driving) to about 100,000 volunteer drivers in the US and Canada. And the media is full of stories about how well self-driving cars perform.

But according to Professor Brown and his team, their actual street performance is a well-kept trade secret that few people know about. So, the researchers conducted an in-depth analysis using 18 hours of YouTube footage filmed by enthusiasts testing the cars from the back seat.

One of his video examples shows a family of four standing on the side of a residential street in the United States. There is no pedestrian crossing, but the family would like to cross the road. As the driverless car approaches, it slows down, causing the two adults in the family to wave their hands as a signal to drive the car.

Instead, the car stops right next to them for 11 seconds. Then, as the family begins to walk down the street, the car starts moving again, causing them to jump back onto the sidewalk, after which the person in the back seat rolls down the window and yells, “Sorry. , Self-driving car!”.

“The situation is similar to the main problem we found in our analysis and demonstrates the inability of self-driving cars to understand social interactions in traffic. The driverless vehicle stops so as not to hit pedestrians, but still Driving ends up because it doesn’t understand the signals. Apart from causing confusion and wasting time in traffic, it can also be downright dangerous,” says Professor Brown.

A Drive in Foggy Frisco

In tech-centric San Francisco, the performance of self-driving cars can be closely judged. Here, driverless cars have been abandoned in many parts of the city as buses and taxis, navigating mountain roads amid people and other natural phenomena. And according to the researcher, it has caused considerable backlash among the city’s residents:

“Self-driving cars are causing traffic jams and problems in San Francisco because they react inappropriately to other road users. Recently, the city’s media reported a chaotic traffic jam caused by self-driving cars due to fog. wrote about the phenomenon. Fog caused self-driving cars to overreact, stop and block traffic, even though fog is extremely common in the city,” says Professor Brown.

Robotic cars have been in the works for 10 years and the industry behind them has spent more than DKK 40 billion to further their development. Yet the result has been cars that still drive with many errors, blocking other drivers and disrupting the smooth flow of traffic.

Asked why it is so difficult to program self-driving cars to understand social interactions in traffic, Professor Brown says, “I think part of the answer is that we take the social element for granted . We don’t think about it.” When we get in a car and drive – we do it automatically. But when it comes to designing systems, you need to describe everything we provide and incorporate it into the design. ,

“The car industry could learn from a more sociological approach. Understanding social interactions as part of traffic should be used to design self-driving cars’ interactions with other road users, in a similar way that research has shown the usefulness of mobile phones.” and technology more widely.”

The study is published in the journal Proceedings of the 2023 CHI Conference on the Human Factor in Computing Systems,