One of the most important challenges of autonomous driving systems is the correct relationship to the other occupants of the road. Predicting the behavior of other vehicles is at the center of new research at the MIT Laboratory on artificial intelligence, the CSAIL ( Computer Science and Artificial Intelligence Laboratory ). The task of the AI model developed by the researchers is to classify whether the action of a driver is altruistic or selfish Turn right into a side street or wait until it's your turn? Will there be an intersection or will it try to move, even if it is not your turn? It is hard to know beforehand, but it is also true that all drivers from time to time ̵
Working with people and their environment means guessing their intentions to better understand their behavior. The tendency of people to be collaborative or competitive is often reflected in their driver behavior.
The researchers' intuition was to rely on ideas and concepts developed in the field of social psychology : AI showed 25% more accurate predictions than earlier ones. It will be some time before this technology can be tested on the road – and eventually integrated into "normal" autonomous driving systems – but the stakes are very high: the pedestrian suddenly crossing and getting out of the streak of pedestrians, the killed by Uber's Volvo last year could have survived.