Research On Multi-vehicle Interactive Automatic Driving Lane Change Decision And Planning Based On Game Theory

When changing lanes in a multi-vehicle interaction environment,automatic driving vehicles will interact with multiple vehicles around them in a game,which will affect the decision making and planning thinking.Accurate recognition of the driving style of surrounding vehicles and reasonable planning of their lane change trajectories are the keys to improve vehicle driving safety.In this paper,we analyze the game interactions between autonomous vehicles and surrounding vehicles when changing lanes in a highspeed mixed traffic environment,and construct a humanized decision mechanism and trajectory planning model to realize a reasonable and effective lane change decision and planning,in conjunction with the National Natural Science Foundation of China project "Research on Automatic Driving Situation Awareness and Humanized Robust Decision Planning in High-speed Mixed Traffic Environment"(52072054).The main research is as follows.The main research contents are as follows:(1)A game lane change decision model considering the driving style of surrounding vehicles.In order to construct an accurate vehicle-vehicle interaction game matrix and improve the rationality of the self-vehicle game lane change decision,firstly,the vehicle trajectory data containing lane change behavior are screened based on the NGSIM dataset,appropriate feature indicators are selected,and the noise of feature parameters is removed using symmetric exponential moving average method.Then PCA(Principal Component Analysis,PCA)is used to reduce the dimensionality of the feature parameters to enhance the characterization ability of the feature parameters on vehicle driving style,and Gaussian clustering method is applied to cluster the dimensionality-reduced data into different style categories.Finally,the driving style recognition model based on Support Vector Machine(SVM)is established to realize the driving style recognition of surrounding vehicles.(2)A game lane change decision model considering the driving styles of surrounding vehicles.The dynamic interaction game process and decision mechanism between vehicles in the process of vehicle driving are analyzed.The game theory approach is used to construct the lane change vehicle driving gain function from speed,space and safety features,and the game gain function is optimized by considering the influence of different driving styles on the driving behavior of the lane changing vehicle,and the joint driving style lane change decision model is established.Using SUMO simulation software to set human driving scenarios and mixed driving scenarios,the decision security of Gipps decision model,game decision model and game decision model considering driving styles are compared,and the effectiveness and advantages of the proposed model are proved.(3)Lane change track shape and speed coupling for automatic driving lane change trajectory planning.The trajectory planning problem is divided into two low-dimensional planning problems of track shape planning and velocity planning.For the lane change trajectory planning,the better performance quintic polynomial is selected as the lane change trajectory curve.With the stability of vehicle dynamics as the constraint,a stable trajectory cluster based on the fifth polynomial is determined,a multi-objective optimization function is established,and a genetic algorithm is used to solve the optimal trajectory change.In the longitudinal speed planning,the mathematical graphical representation of the driving environment is established based on the displacement time S-T diagram,and the speed profile satisfying the safety collision avoidance constraints and vehicle dynamics constraints is searched in the S-T diagram using the hybrid A*algorithm,and the optimal speed planning is obtained using the multi-objective function and quadratic planning optimization.Finally,the simulation verifies the effectiveness of the lane change trajectory planning method.