Human-Like Decision-Making Based on Sequential Games for Automated Vehicles Considering Subjective Cognition
Uncontrolled intersections are highly dynamic and strongly interactive decision-making scenari-os,in which it is a challenging task to enable automated vehicles to make safe and reasonable decisions similar to skilled drivers and pass through the intersections successfully.The subjective attributes of ontology in cognition and decision-making process are fully considered in this paper,and an interactive human-like decision-making method based on sequential games for automated vehicles is proposed.Firstly,the multi-objective driving triggers are deep-ly explored from multiple dimensions such as traffic efficiency,space margin,ride experience,and driving safety.Further,a game decision-making model is established,which is embedded with personalized and human-like driv-ing characteristics and can match driver and passenger groups with different driving modes and types.On this basis,the concept of sequential priority and the self-perspective decision-making scheme that imitates human logic are pro-posed to realize self-evolution of sequential patterns of rolling stage game decision-making.Finally,the effective-ness of the proposed method is verified through multiple sets of comparative experiments.The results show that the interactive human-like decision-making method proposed in this paper can resolve potential conflicts and deal with safety decision-making problems in a continuous and interactive manner,while improving the naturalized and hu-man-like effect of personalized decision-making of automated vehicles.
万方数据
维普
