| In the 1980s,private cars began to appear in China,and car ownership continued to grow with the increase of national income.By 2017,the number of private cars in China has reached a staggering 180 million.With the increasing number of cars,people travel more convenient,life more convenient.However,the increase of the number of cars has also led to a series of problems,such as environmental pollution,traffic safety,traffic jam and so on.Most of the accidents are related to the driver.In order to improve the safety of driving,people begin to imagine whether the vehicle can drive independently without relying on the driver.In the above background,the focus of this topic is the autonomous driving lateral control,and some research has been carried out.First,this paper analyzes the research status of autonomous driving and lateral control.Aiming at the problems existing in the research,the main research content of this topic is the model-free lateral control.Second,the advantages and disadvantages of classical control,modern control and intelligent control are analyzed and compared.Combined with the research direction of this subject,the optimal control of modern control methods is selected as the control method of this study.Third,the simplified linear vehicle model is established.For the unknown system model,the reinforcement learning theory is introduced,and the model-free reinforcement learning method is used to learn vehicle system model knowledge spontaneously to obtain the optimal control of the system.In this paper,model-free reinforcement learning optimal control method is proposed,and compared with the traditional method,it is proved that the method has good convergence.Fourth,considering the passenger comfort and fuel economy,the input is designed as a saturation function in this study.Combined with the model-free reinforcement learning control,a low gain state feedback control design method for lateral control is proposed..The feasibility of this method is proved by Matlab simulation test.Fifth,considering the influence of driver control input,the dynamic game theory is adopted.Under the premise of unknown model,combined with reinforcement learning and game theory,a lateral control strategy based on lane keeping of driver model is proposed,in which driver and vehicle control module work together,and the experiment shows that the method is stable. |
