Game Behavioral Decision-Making Mechanism And Multi-Objective Trajectory Planning For Lane Changing Of Autonomous Driving

Autonomous vehicles have the advantages of fast response and high work efficiency,which provides new ideas and methods for effectively alleviating traffic problems.Lane-changing behavior is one of the frequent driving behaviors.Because of its complexity,it is easy to induce traffic accidents and cause traffic congestion.At present,autonomous vehicles are showing a trend of popularization,and it is urgent to improve the lane-changing ability of autonomous vehicles.Therefore,the research direction focuses on lane-changing behavioral decision-making and trajectory planning in dynamic traffic scenes,and deeply analyzes the lane-changing mechanism of autonomous vehicles,in order to promote the improvement of autonomous lane-changing ability and traffic problems.Firstly,the typical application scenes of autonomous driving are clarified.The personalized functions of key technologies for autonomous vehicles and their synergistic relationship in completing driving tasks are expounded.The intelligent decision-making planning module is further explored,and the logic of formulating lane-changing behavioral decision-making and the process of planning the lane-changing trajectory are integrated to form the framework of lane-changing behavioral decision-making and trajectory planning,which provides a theoretical basis for the advancement of research content.Secondly,explain the game lane-changing phenomenon of autonomous driving,and analyze lane-changing strategies and game gains.Moreover,objectively quantify lane-changing intention,and introduce the lane-changing collision probability and lane-changing dynamic risky coefficient,so as to construct the decision-making behavioral model of lane-changing game.The SUMO platform is selected for lane-changing experiments.The decision-making behavioral model of lane-changing game is compared with the traditional lane-changing model to verify the rationality of game lane-changing theory.Then,from the perspective of interaction,the research on lane-changing behavioral decision-making is promoted from the game characteristics to the internal mechanism of molecular dynamics.The system similarity analysis of vehicles and molecules is carried out,and the interaction potential is introduced to construct the molecular dynamics lane-changing model,revealing the essence of interactive dynamic lane-changing of vehicles on the longitudinal and lateral two-dimensional planes.The SUMO platform is used for lane-changing experiments.The performance of the decision-making behavioral model of lane-changing game and the molecular dynamics lane-changing model is compared and analyzed,which verifies the scientificity of the molecular dynamics lane-changing decision-making mechanism.Finally,the research on lane-changing of autonomous vehicles are extended from the behavior decision-making layer to the local path planning layer,and the critical collision conditions in different lane-changing scenes are analyzed.Besides,considering the comfort requirements and boundary characteristics of lane changing,the multi-objective constraint conditions of lane-changing trajectory are formed,and the lane-changing trajectory planning model in multiple scenes is constructed.Using MATLAB to give practical significance to the lane-changing trajectory planning model.The results show that the autonomous vehicles can avoid the lane-changing risk.Besides,the lane-changing trajectory is smooth and continuous,which meets the multi-objective requirements of the lane-changing process.