Intent prediction of vulnerable road users for trusted autonomous vehicles

This study investigated how future autonomous vehicles could be further trusted by vulnerable road users (such as pedestrians and cyclists) that they would be interacting with in urban traffic environments. It focused on understanding the behaviours of such road users on a deeper level by predicting their future intentions based solely on vehicle-based sensors and AI techniques. The findings showed that personal/body language attributes of vulnerable road users besides their past motion trajectories and physics attributes in the environment led to more accurate predictions about their intended actions.