Research On Path Tracking Control Method Based On Model Predictive Control

Path tracking control is one of the key technologies for intelligent vehicles to realize self-driving.However,the complex coupling relationship between longitudinal and lateral motions has great limit on the development of path tracking control on intelligent vehicle.The independent longitudinal control and lateral control cannot meet the requirements for driving on extreme conditions such as high-speed cornering and low-adherent road.In addition,sensitivity of path tracking control performance of system parameter is the main reason for its low portability.On account of this,a preview MPC path tracking control method based on rotation angle compensation was proposed by combining with compensation control ideas.,which was based on the Model Predictive Control(MPC)principle and vision preview theory.The method was aiming at improving the tracking performance of the path tracking system.The main research content is as follows:Firstly,a six-degree-of-freedom dynamics model and the tire model were established,and the range and the necessity of side angle constraint at the condition of stable driving were analyzed based on the slip characteristic curve of the tire model.Based on this,the system dynamics model was further simplified,and the path tracking control method of Linear TimeVarying Model Predictive Control(LTV MPC)based on coordinate transformation was studied.The control horizon and prediction horizon affecting the system tracking performance were simulated and analyzed to conclude that the control method is more dependent on the horizon parameters.Therefore,the advantages and disadvantages of the visual preview control were analyzed.the preview MPC path tracking control method was proposed by combining the preview control and the MPC.The lateral dynamics had little influence on longitudinal dynamics,and the selection of the preview distance only depended on the longitudinal speed and the desired path.On account of this,a separate longitudinal speed tracking control strategy was set up while the longitudinal acceleration and the preview distance were considered as the fixed value in each control horizon.As a result,the multi-input and multioutput system was converted into a single-input and multi-output system to achieve integrated lateral and longitudinal control for intelligent vehicle.In addition,the longitudinal expected speed model was established by taking vehicle speed as another control volume besides the steering angle to keep the variable speed track of the desired path.At the same time,tracking controller characteristics can’t be optimized because the preview distance can’t fully reflect the changes of variables.On account of this,the steering angle compensation controller based on the compensation control was designed,and the online adaptive adjustment of the compensation control parameters was realized based on the fuzzy control theory.Finally,simulation comparison analysis and experimental verification of the designed preview tracking MPC path tracking controller based on rotation angle compensation were carried out.The results showed that the control method proposed can meet the driving requirements of different working conditions without changing the system parameters in the path tracking process.The peak value of lateral error was reduced by 26.3%,and the peak value of direction error was reduced by 8.7% when comparing with the traditional MPC control algorithm at speed of 54 km/h.It can quickly eliminate the overshoot of lateral error caused by large curvature path tracking at speed of 72 km/h.Furthermore,the path tracking performance was further improved by combining variable speed control strategy to be superior to the traditional MPC path tracking control.