When crossing the road, pedestrians tend to make eye contact with the driver in order to get confirmation that they are seen. What if it’s self-driving cars? The developers propose to install artificial eyes on the machines.
Scientists at the University of Tokyo conducted their own study by equipping a golf cart with artificial human-like robotic eyes. The driver was put into the cabin, hidden behind a mirror film on the windshield, which created the feeling that there was no person in the car who was driving it.
During the experiment, four videos were shot, in which the camera recorded the moments when a golf cart approached a pedestrian crossing. In the first video, the car stopped “looking” at the pedestrian, in the second, it ignored the person with its eyes and drove by without stopping. In two more commercials, the golf cart performed the same actions, but it no longer had robotic eyes.
18 adult volunteers were also selected for the study – 9 men and 9 women. They watched previously filmed videos in which, using a virtual reality headset, they saw the environment from the point of view of pedestrians. The videos were shown several times in random order, and the subjects had to decide within 3 seconds whether it was safe to cross the road in each of the situations.
The study found that men were more likely to cross when it was unsafe (i.e. when the golf cart didn’t look like it was about to stop), while women were more likely not to cross when it was safe to do so (i.e. when the car is stopped). However, both errors occurred less frequently when the cart was equipped with eyes: men reported that the situation seemed more risky when the eyes looked the other way, while women reported feeling more secure when the eyes looked at them.
The researchers say the results will be different if they run a larger experiment using more scenarios and participants. However, they still believe that the “artificial gaze” embedded in self-driving cars is a great idea.
“If robotic eyes on self-driving cars can really contribute to safety and reduce traffic accidents, we should seriously consider implementing them. In the future, we would like to develop automatic control of robotic eyes connected to AI for self-driving, which can adapt to different situations,” said one of the authors of the study, Professor Takeo Igarashi.
The research paper was presented at the 14th International Conference on Automotive User Interfaces and Interactive Automotive Applications. However, this is not the first attempt to make traffic safer on an unmanned vehicle – in 2018 Jaguar Land Rover conducted a test that proved that pedestrians felt safer crossing the road in front of special automated “modules” with moving eyes that “followed” them.
Source: Newatlas