Dynamic decision making in lane change: Game theory with receding horizon

Abstract

Decision making for lane change manoeuvre is of practical importance to guarantee a smooth, efficient and safe operation for autonomous driving. It is, however, challenging. On one hand, the behaviours of ego vehicle and adjacent vehicles are dependent and interactive. On the other hand, the decision should strictly guarantee safety during the whole process of lane change with uncertain and incomplete information in a dynamic and cluttered environment. To this end, the concept of Receding Horizon Control (RHC) is integrated into game theory in conjunction with reachability analysis tool, resulting in RHC based game theory. Specifically, the decision of each game relies on not only uncertain information at current step but also the future information calculated by reachability analysis. The decision is repeatedly made with the advent of new information using the concept of RHC. As a result, safety can be guaranteed during the whole process of lane change in a dynamic environment. Case study is conducted to demonstrate the advantages of the proposed approach. It is shown that the proposed RHC based game theory approach incorporating uncertain information can provide a safer and real-time decision.