The fatal accident involving a self-driving vehicle in Arizona has researchers busy trying to get the technology to better recognize pedestrians.
Autonomous vehicles are good at recognizing mechanical objects like cars and trucks. But people are more difficult to detect. And their path of travel is even harder to predict.
Researchers at the University of Waterloo, in Ontario, cite many variations that human’s pose to the recognition software. For example, groups of people crossing a street look different than one person. Sometimes, pedestrians may use a walker, or have a bicycle.
Self-driving researchers say the accident in which an Uber autonomous vehicle ran over and killed a woman pedestrian in Arizona will prompt developers to redouble their efforts to make the technology safer.