MSGID: 1:317/3 6476cdb7   
   PID: hpt/lnx 1.9.0-cur 2019-01-08   
   TID: hpt/lnx 1.9.0-cur 2019-01-08   
    Self-driving cars lack social intelligence in traffic    
      
     Date:   
         May 30, 2023   
     Source:   
         University of Copenhagen - Faculty of Science   
     Summary:   
         Self-driving cars fall short when it comes to understanding the   
         social codes in traffic that let human drivers decide whether to   
         give way or drive on, according to new research.   
      
      
         Facebook Twitter Pinterest LinkedIN Email   
      
   ==========================================================================   
   FULL STORY   
   ==========================================================================   
   Should I go or give way? It is one of the most basic questions in traffic,   
   whether merging in on a motorway or at the door of the metro. The decision   
   is one that humans typically make quickly and intuitively, because doing   
   so relies on social interactions trained from the time we begin to walk.   
      
   Self-driving cars on the other hand, which are already on the road in   
   several parts of the world, still struggle when navigating these social   
   interactions in traffic. This has been demonstrated in new research   
   conducted at the University of Copenhagen's Department of Computer   
   Science. Researchers analyzed an array of videos uploaded by YouTube   
   users of self-driving cars in various traffic situations. The results   
   show that self-driving cars have a particularly tough time understanding   
   when to 'yield' -- when to give way and when to drive on.   
      
   "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. 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," says Professor Barry   
   Brown, who has studied the evolution of self-driving car road behavior   
   for the past five years.   
      
   Sorry, it's a self-driving car!  Companies like Waymo and Cruise have   
   launched taxi services with self-driving cars in parts of the United   
   States. Tesla has rolled out their FSD model (full self-driving) to   
   about 100,000 volunteer drivers in the US and Canada. And the media   
   is brimming with stories about how good self-driving cars perform. But   
   according to Professor Brown and his team, their actual road performance   
   is a well-kept trade secret that very few have insight into. Therefore,   
   the researchers performed in-depth analyses using 18 hours of YouTube   
   footage filmed by enthusiasts testing cars from the back seat.   
      
   One of their video examples shows a family of four standing by the curb   
   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, 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 the family begins walking across the road, the car starts moving   
   again, causing them to jump back onto the sidewalk, whereupon 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 to not   
   hit pedestrians, but ends up driving into them anyway because it doesn't   
   understand the signals. Besides creating confusion and wasted 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 judged up close. Here, driverless cars   
   have been unleashed in several parts of the city as buses and taxis,   
   navigating the hilly streets among people and other natural phenomena. And   
   according to the researcher, this has created plenty of resistance 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 wrote of a chaotic traffic event   
   caused by self-driving cars due to fog.   
      
   Fog caused the 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 over DKK 40 billion to push their development. Yet the   
   outcome has been cars that still drive with many mistakes, blocking   
   other drivers and disrupting the smooth flow of traffic.   
      
   Why do you think it's so difficult to program self-driving cars to   
   understand social interactions in traffic?  "I think that part of   
   the answer is that 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. The car industry could learn from having a more sociological   
   approach. Understanding social interactions that are part of traffic   
   should be used to design self- driving cars' interactions with other   
   road users, similar to how research has helped improve the usability of   
   mobile phones and technology more broadly."  About the study:   
       * The researchers analyzed 18 hours of video footage of self-driving   
       cars   
         from 70 different YouTube videos.   
      
       * Using different video analysis techniques the researchers studied   
       the   
         video sequences in depth, rather than making a broader superficial   
         analysis.   
      
       * The study is called: "The Halting Problem: Video analysis of   
       self-driving   
         cars in traffic" has just been presented at the 2023 CHI Conference   
         on Human Factors in Computing Systems, where it won the conference's   
         best paper award.   
      
       * The study was conducted by Barry Brown of the University of   
       Copenhagen   
         and Stockholm University, Mathias Broth of Linko"ping University,   
         and Erik Vinkhuyzen of Kings College, London.   
      
       * RELATED_TOPICS   
             o Matter_&_Energy   
                   # Transportation_Science # Automotive_and_Transportation #   
                   Virtual_Environment # Aerospace   
             o Computers_&_Math   
                   # Computers_and_Internet # Video_Games #   
                   Distributed_Computing # Computer_Modeling   
       * RELATED_TERMS   
             o Traffic_engineering_(transportation) o   
             Alternative_fuel_vehicle o Road-traffic_safety o   
             Positron_emission_tomography o Alcohol_fuel o Science o   
             Supercomputer o Flexible-fuel_vehicle   
      
   ==========================================================================   
   Story Source: Materials provided by   
   University_of_Copenhagen_-_Faculty_of_Science. Note: Content may be   
   edited for style and length.   
      
      
   ==========================================================================   
      
      
   Link to news story:   
   https://www.sciencedaily.com/releases/2023/05/230530125438.htm   
      
   --- up 1 year, 13 weeks, 1 day, 10 hours, 50 minutes   
    * Origin: -=> Castle Rock BBS <=- Now Husky HPT Powered! (1:317/3)   
   SEEN-BY: 15/0 106/201 114/705 123/120 153/7715 218/700 226/30 227/114   
   SEEN-BY: 229/110 112 113 307 317 400 426 428 470 664 700 291/111 292/854   
   SEEN-BY: 298/25 305/3 317/3 320/219 396/45   
   PATH: 317/3 229/426