A Game-Based Hierarchical Model for Mandatory Lane Change of Autonomous Vehicles

Ping Huang, Haitao Ding, Zhenjia Sun, Hong Chen

Abstract: Game theory-based decision-making model provides an effective means to enable intelligent and human-like Mandatory Lane Change (MLC), which is closely linked to driving safety and efficiency. However, current relevant models have limitations, such as imperfect game structure and incomplete information considered in payoff definitions, with the root cause of ignoring differences in driving styles between interacting vehicles, which are directly related to the acceptable safety thresholds of drivers. To address this issue, this study presents a novel game theory-based decision-making strategy, considering diverse driving styles, achieved by constructing a game with a variable structure according to the Relative Driving Style (RDS) between vehicles. Validation of the Next Generation SIMulation (NGSIM) dataset shows that the proposed decision-making strategy achieves an average accuracy of 98%, which is superior to that of existing single-type game theory-based algorithms.