A study has found that autonomous vehicles cause traffic congestion with a risk of being hazardous because they fail to understand human behavior. The study shows how self-driving cars cannot comprehend non-verbal communication crucial in driving decisions. One instance is when deciding to drive on or hold back in traffic. This is one of the decisions motorists make in a split second. Autonomous cars cannot pick up specific cues in traffic; thus, they end up causing congestion.
Barry Brown, a professor from the University of Copenhagen, explains how traffic navigation is not solemnly dependent on traffic rules but on other factors such as body language. Programmers need to capture this aspect when it comes to self-driving cars. It can be infuriating and a peril to road users. Despite various companies deploying autonomous vehicles in the United States and Canada, Brown argues that the performance of such cars is a trade secret concealed from the public eye.
Videos showing self-driving cars being a hazard to pedestrians
There have been many incidences of autonomous cars being a danger. In one instance, a family in the US tries to cross a street. One thing noted is that the street doesn’t have a crosswalk. However, the family still attempts to cross the road. They first gesture the autonomous car to proceed, but it stops for 11 seconds. The family takes this as a sign to cross and starts across, only for the vehicle to start moving again. The family hurriedly leaps back.
Brown explains how the autonomous car fails to pick up the cues and interpret them. It almost runs into them even after discovering they are pedestrians, and it stops initially. He suggests that nonverbal communication should be included in designing of self-driving cars.
How manufacturers can fox the issue with traffic signals
There has been a suggestion of introducing a fourth light which will give self-driving cars a chance to control traffic whereas motorists are alerted on what’s happening. The other traffic lights will still perform the same function. Professor Hajbabaie insists that distributed computing will bring efficiency to autonomous driving using their computing resources.
