Multi-objective Optimal Human-like Driving Trajectory Planning Considering Scenarios Spatiotemporal Characteristics

In order to make full use of spatiotemporal characteristics of driving scenarios, make trajectory planning meet various driving needs, and obtain safe and stable human-like driving trajectories, a multi objective optimal human-like driving trajectory planning model considering the spatiotemporal characteristics of scenarios is constructed. Based on the graph theory method, the global spatiotemporal interaction feature fusion method of dynamic driving scenarios is studied. According to the theory of imitation learning, a human-like driving trajectory planning network model considering the spatiotemporal characteristics and the influence of other vehicles’ predicted trajectories is constructed. For those multiple candidate planning trajectories output by the network, the idea of multi-objective optimization is used to comprehensively evaluate and select them, and then the multi-objective optimal human-like driving trajectory in dynamic scenarios is obtained. The proposed trajectory planning method is tested and verified by simulation experiments. The experimental results show that the planning model can efficiently handle the spatiotemporal interaction characteristics of driving scenarios, make accurate judgments and real-time responses to different driving behaviors of other traffic participants, and output multi-objective optimal human-like driving planning trajectories.