Decision-making And Control Method For Connected Autonomous Multi-vehicle Driving Behavior Based On Game Theory

Scientific,technological progress and the development of society have made mankind's requirements for vehicle safety,efficiency,and comfort flourish.With the development of microelectronics,computers,automatic intelligent technology,artificial intelligence,and VANET,connected autonomous vehicles with high-level autonomous driving function combined with 5G-V2 X technology will gradually realize large-scale commercial applications,and become a smart transportation system and improve the key to the road safety levels.The decision of vehicle driving behavior is an important factor to ensure traffic safety and improve driving efficiency.Therefore,this thesis chooses the game theory method to predict the vehicle behavior,and combines the control technology to verify the feasibility of the strategy.Initially,construct a two-degree-of-freedom vehicle dynamics model and driver model;calculate the safe distance between vehicles from the perspective of decision-making planning,and explore the different responses and incentive factors for stimulating lane selection;select the fifth-order polynomial lane change trajectory for trajectory prediction.Secondly,considering the influence of vehicle driving strategy selection based on the interaction of surrounding vehicle behaviors,the extensive-form-game trees' research method is used to decompose the multi-vehicle behavior interaction into two or more dynamic cooperative games based on the sequence of actions under perfect information.The inductive method solves the perfect Nash equilibrium of the subgame,and obtains the optimal strategy based on the interdependence of the workshop,and calculates the corresponding payoffs.Furthermore,based on the optimal strategy combined with the vehicle dynamics model,the driver model,and the vehicle of lane change trajectory model,a classic PID longitudinal control strategy and an upper-level control system based on feedforward and feedback for optimal lateral path following are completed.Eventually,the pre-imitation platform of PreScan and Matlab / Simulink is built,and the optimal decision-making and control algorithm are used to simulate the behavior effect of multi-vehicle games under the two conditions of urban area and highway.Experiments show that the subject vehicle chooses the left lane change behavior,and the surrounding vehicles give the corresponding behavior status changes,and finally maintain a reasonable expected return.