Safe, Efficient and Socially-Compatible Decision of Automated Vehicles: A Case Study of Unsignalized Intersection Driving

Safe and smooth interaction between other vehicles is one of the ultimate goals driving automation. However, recent reports demonstrative deployments automated (AVs) indicate that AVs are still difficult to meet expectation interacting drivers, which leads several AV accidents involving human-driven (HVs) without understanding about dynamic process. By investigating 4300 video clips traffic accidents, it found limited visual field drivers leading factor in inter-vehicle accidents. A game-theoretic decision algorithm considering social compatibility proposed handle with a truck at an unsignalized intersection. Starting from probabilistic model for characteristics fitness reciprocal altruism incorporated game payoff design. Human-in-the-loop experiments carried out, 24 subjects invited drive interact deployed two comparison algorithms. Totally, 207 cases intersection interactions obtained analyzed, shows decision-making can improve both safety time efficiency, make decisions more line human drivers. These findings help inform design algorithms, ensure be safely efficiently integrated into human-dominated traffic.