Research On Human-machine Co-driving Control Based On Model Prediction And Game Theory

Since the driver is the main factor that leads to frequent traffic accidents,and the autonomous driving technology is still not mature enough,the human-machine mutual driving control is a necessary transitional phase from assisted driving to autonomous driving.The four wheels of the four-wheel independently driving electric vehicles are controlled independently,and their intelligent wire-controlled chassis contains multiple driving assistance systems,so they are ideal for carriers that are controlled by humans and machines.Therefore,this paper takes the four-wheel independently driving electric vehicles as the research object.Aiming at the problem of human-machine mutual driving control strategy when realizing path tracking and stability control,related researches are carried out based on model predictive control theory and game theory.The main research work is as follows:(1)Establishment of the four-wheel independently driving electric vehicle model.Car Sim and MATLAB/Simulink are used to transform the traditional prototype C-Class Car into an electric vehicle,which can realize active front/rear wheel steering and four-wheel independently driving.The vehicle model is simulated and verified.The comparison results show that all important indicators of the established electric vehicle model are very close to the prototype vehicle,and the accuracy is very high,which lays the foundation for the subsequent verification of the effectiveness of the co-driving controller.(2)Research on the driver-AFS co-driving control strategy based on centralized model predictive control.In response to the path tracking and stability control of human-machine co-driving intelligent vehicles under the mutual control of the driver and AFS,a driver-AFS co-driving controller based on centralized model prediction control theory is proposed.A two-degree-of-freedom vehicle model is established as the predictive model.The corresponding objective function is established,with the limit steering angle of the front wheel as the constraint,and tracking the desired lateral displacement,yaw angle and yaw rate as the goal.The quadratic programming algorithm is used to solve the constrained optimization problem,and the optimal steering angles of the driver and AFS are obtained respectively to control the vehicle together.The simulation results show that the designed centralized MPC co-driving controller can coordinate the driver and AFS to jointly control the vehicle to track the target path accurately and stably under severe driving conditions,especially the stability control effect.Compared with LQR control algorithm,its yaw rate error is reduced by 43.08% and 24.24% under two different driving conditions.(3)Research on the driver-AFS-ARS co-driving control strategy based on non-cooperative Nash game.The distributed differential game control framework can effectively describe the mapping relationship between different targets and different actuators.Therefore,the driver-AFS-ARS co-driving controller is designed based on the non-cooperative game theory.The first two take tracking the path as the goal,and the latter takes the control stability as the goal.The three are coordinated and controlled by game theory,and the distributed model predictive control method is used to solve the Nash equilibrium.The simulation results show that the proposed non-cooperative game-based driver-AFS-ARS co-driving control strategy can effectively track the target path and maintain vehicle stability.Compared with the driver-AFS game control,the lateral speed error is reduced by 83.04% and 68.54%,respectively,and the yaw rate error is reduced by 70.42% and 39.67%,respectively,under the constant speed and acceleration conditions,which greatly improves the active safety of vehicle driving.(4)Research on the driver-AFS-DYC hierarchical control strategy based on non-cooperative Nash game.Under extreme conditions,the stability control margin of DYC is greater than that of ARS.Therefore,a hierarchical control strategy is proposed for the path tracking and stability control problems of human-machine co-driving vehicles.The upper layer designs the driver-AFS-DYC co-driving controller based on the non-cooperative Nash game,and uses the distributed model predictive control method to solve the Nash equilibrium.The lower layer designs a torque distribution controller based on the sequential quadratic programming algorithm.With the goal of minimizing tire utilization,the virtual additional yaw moment output of the upper layer is distributed to each in-wheel motor.The simulation results show that the driver-AFS-DYC co-driving controller can ensure the stability of the vehicle while accurately tracking the target path.Compared with the driver-AFS game controller,its stability effect has been greatly improved.The specific performance is: under two different simulation conditions,the yaw rate is improved by 50.09% and 53.09% respectively.