Application Of Model Predictive Control In Trajectory Planning And Vehicle Control

As for the research of automatic driving more and more deeply,now the landing of it generally adopts hierarchical solutions.This means that the system will be transformed into some small subsystems,in which the trajectory planning and tracking control subsystems occupy a very important position,so it is of great significance to study them.And this paper starts to construct the vehicle lateral dynamics model and longitudinal inverse model.And then,we studied on the vehicle lane keeping,fast turning and minimum lap time based on three predictive control methods.The main research contents are as follows:1.Firstly,in order to meet the demand of the predictive control,the lateral deviation model and the longitudinal inverse model of the vehicle are constructed.In detail,the vehicle lateral deviation model adopts the bicycle model and in condition of the vehicle motion geometry,the kinematic model of the vehicle is constructed first,and then the dynamic model is constructed through the approximate relationship between the tire lateral force and the sideslip angle at a small angle.Next,the vehicle longitudinal inverse model is prepared for the following longitudinal speed tracking.Firstly,according to the positive and negative logic of the desired acceleration,the accelerating mode or braking mode is selected.Then,through the torque relationship,transmission relationship and other contents,the vehicle longitudinal control inverse models of the two modes are constructed.2.In order to achieve the lane keeping,a method based on the predictive function control is designed.This method constructs the consistent point of the prediction through the lateral deviation model of the vehicle,and constructs the whole solution process of the predictive functional control method through the reasonable attenuation coefficient,and then carries out the experimental verification through the joint simulation of Matlab/Simulink and CarSim.Next,in order to realize the fast turning,a method based on curve interpolation and generalized predictive control is designed.Firstly,the discrete points of the desired trajectory are obtained by the numerical optimization,then the desired trajectory is fitted by the curve interpolation method,and then the desired trajectory is tracked by the trajectory tracker based on the GPC to achieve the desired effect.Besides,in order to realize the longitudinal speed tracking,a speed tracking controller combining predictive functional control with inverse model is designed.In this method,a predictive functional controller is added before the longitudinal inverse model,which makes up for the inaccuracy of the longitudinal inverse model.3.In order to achieve the shortest cycle time,a method of shortest cycle trajectory planning based on model prediction and a method of trajectory tracker based on dual-mode model predictive control are proposed.Firstly,the real-time trajectory planning is constructed by vehicle kinematics model prediction.Then considering the calculation time,some trade-off improvements are made to solve the optimization problem of online planning.These improvements include the initial guess in numerical solution,the selection of prediction step and the selection of the construction of the optimization method.Finally,the reference trajectory is tracked by dual-mode model predictive control.