Research On Autonomous Vehicle Lane Change Based On Game Theory

In order to improve the autonomous lane change performance of driverless vehicles,this paper presents an in-depth analysis of the current lane change strategy and problems in the lane change process of driverless vehicles,and establishes an autonomous lane change strategy based on game theory,combined with an improved support vector machine(SSA-SVM)method.After the vehicle makes the lane-changing decision,to ensure that the vehicle can follow the reference trajectory,this paper adopts a five-degree polynomial lane-changing trajectory and adds lateral acceleration constraints to improve the stability of the vehicle when changing lanes at high speed.The vehicle is modelled using three-degree-of-freedom vehicle dynamics and an algorithm for a transverse-longitudinal dual-model predictive controller is designed to control the lateral displacement and longitudinal velocity of the vehicle.A joint simulation study was carried out and analysed using the vehicle dynamics software Carsim,the scenario building software Prescan and the simulation platform software Matlab.The results show that the method proposed in this paper can effectively improve the autonomous lane changing capability of the vehicle:(1)In order to provide normative data for machine learning,this paper extracts lane change data from the real driving dataset NGSIM.Due to the existence of certain errors in the dataset itself,in order to improve the accuracy of the later model training,this paper adopts the symmetric exponential moving average method(s EMA)to smooth the data and extracts the vehicle lane change trajectory according to the formulated rules.(2)Construct a game decision model for the vehicle driving process,study the game relationship between the vehicle changing lanes and the surrounding vehicles,and give a coping strategy based on full information without collaboration.In order to obtain a more objective game theory gain matrix,a support vector machine(SSA-SVM)based on the sparrow search algorithm is designed,and the probability of executing the lane change decision by the improved support vector machine is used as the game gain matrix,so as to improve the accuracy when making the vehicle decision.(3)Establish a vehicle lane change trajectory tracking model.For the three-degree-of-freedom vehicle dynamics model of the vehicle,a lane change model based on a five-degree-of-freedom polynomial trajectory is proposed,adding a vehicle lateral acceleration constraint,and a model prediction controller based on the three-degree-of-freedom vehicle dynamics is designed to convert it into a quadratic programming problem and optimise it by adding a type lateral deflection angle constraint to improve the stability of the vehicle during high-speed driving.(4)Simulation of vehicle lane change working conditions.Using Carsim,Prescan and Matlab software to carry out joint simulation experiments to verify the trajectory planning results under the lane change working conditions,the working conditions of the lane change model are scenario-built,and the algorithm simulation is realized by transmitting the surrounding traffic flow information to the simulink model through the sensors,and the results show that the improved game theory based lane change decision model proposed in this paper under the highway working conditions The results show that the improved game theory-based lane change decision model proposed in this paper performs well under highway conditions.