Design Of Lateral Controller For Intelligent Driving Vehicle Based On Model Predictive Control

With the development of artificial intelligence and hardware level,autonomous vehicle have gradually become one of the research hot spots of universities and technology companies.As one of the core technologies of autonomous vehicle,vehicle motion control technology has an important position.And one of the important components of vehicle motion control technology is the vehicle's lateral control technology.This paper introduces the model predictive controller based on the vehicle kinematics model and the vehicle dynamics model respectively as the lateral controller of the vehicle,and introduces the exponential decay weight matrix to improve the controller's control performance.To solve the problem of the model prediction controller large amount of calculation,the Laguerre function is introduced to optimize the calculation process,and improve the real-time performance of the controller.This paper introduces the principle of the model predictive controller,and analyzes the tracking ability and stability of the system under the model predictive controller based on the vehicle kinematics and dynamics models.Analyze and compare the tracking ability and stability of the system bewteen the two controllers.The controller of the vehicle dynamics model takes into account the dynamic characteristics of the vehicle,and the tracking capability of the system is significantly better than that of the controller based on the vehicle kinematics model at high and medium speeds.Simulate and analyze the effects of different parameters of the model prediction controller on the system tracking performance and calculation time.Because of the contradiction between the controller's response speed and stability,an exponential decay weight matrix is introduced into the cost function.By redistributing the weights in the prediction time domain and reducing the weights of future errors and control variables,the goals,optimizing the response speed of the controller and improvming the tracking performance of the system without affecting the stability of the system are achieved.At the same time,the introduction of exponential decay weight is also helpful to reduce the numerical problems caused by too long prediction time domain.Then for the problem of poor real-time controller and large amount of calculation,the Laguerre function is introduced.The simulation time comparison of the model predictive controller shows that the influence of the control time domain on the calculation amount far exceeds the prediction time domain.Therefore,similarly to system identification,the Laguerre function is used to fit the control sequence,and the control sequence can be expressed with a smaller number of items,so as to reduce the optimization dimension of the controller calculation process and optimize the real-time performance of the controller.