Research On Tracking Control Of Autonomous Vehicle Based On Model Predictive Control

With the rapid development of computer technology,chip manufacturing and communication technology,autonomous driving technology is widely used in port logistics,unmanned mining areas,urban roads and other scenes.It mainly includes three technical fields:environmental perception,trajectory planning and tracking control.Among them,tracking control is used to generate smooth control signals to ensure passenger comfort and tracking accuracy,which is one of the basic problems of autonomous driving technology.In addition,vehicle dynamics has the characteristics of nonlinearity,parameter uncertainty,velocity-varying,etc.In the actual vehicle system,there is control input delay between the upper controller and the actuator through the steering-by-wire(SBW)system.Ignoring the influence of dynamic characteristics and the SBW system will lead to a decrease in vehicle tracking accuracy,and steering wheel shaking and other issues may occur when tracking large curvature path.Therefore,the research for tracking control strategy of both dynamic characteristics and SBW system in scenarios with large curvature path has significant research value and practical significance.The main research contents of this paper are as follows:Firstly,in order to improve the tracking accuracy of the autonomous driving vehicle on the curve path,an improved kinematic model with look-ahead distance and curvature simultaneous is developed.The tracking error of the look-ahead point can be calculated and the change of path curvature can be perceived in advance by the look-ahead distance.A nonlinear model predictive controller is developed to maintaining the vehicle stability and minimizing the lateral trajectory tracking error,where path curvature is considered as an input constraint to improve driving comfort.The iterative feasibility and stability of the proposed controller are analyzed.The experimental results show that the proposed controller can effectively guarantee the stability and tracking accuracy of the autonomous vehicle.Then,the path tracking problem under the condition of velocity-varying and control input delay is considered.Considering the velocity as a time-varying parameter,and modeling the input delay problem,a vehicle lateral dynamics model based on linear parameter varying(LPV)is established.A path tracking controller based on robust model predictive control is designed which can guarantee the vehicle stability and improve the tracking accuracy under the condition of velocity-varying and input delay.The control signal is obtained by solving the optimization problem based on linear matrix inequality(LMI).The control system is evaluated and validated by MATLAB simulation.Finally,a path tracking controller based on robust model predictive control and H_∞is proposed for autonomous vehicles under condition of model uncertainties and external disturbances.Considering the uncertainty of tire cornering stiffness and the product of longitudinal velocity and curvature as external disturbance,the vehicle lateral dynamics model is improved by LPV approach.In order to satisfy the Lyapunov asymptotic stability and H_∞performance constraints,the controller obtains the control signal by solving the optimization problem based on LMI,and improves the real-time computing efficiency by offline computing.The simulation results by MATLAB show that the controller has strong robustness while ensuring the tracking accuracy.