Data-driven Model-free Adaptive Overtaking Control On Highway

Due to the booming development of robotics and sensing technology,unmanned driving has gradually become a research hotspot at home and abroad.Automatic overtaking on expressway is a typical driverless behavior,which can easily cause traffic accidents.The existing automatic overtaking control methods are mostly based on models and have poor portability.Therefore,it is of great theoretical significance and practical application value to study the data driven automatic overtaking control method which is not dependent on the model and has stronger portability.Model Free Adaptive Control(MFAC)is a typical data-driven control algorithm.It does not rely on the precise mathematical model of the system,but only uses the input and output data to complete the control of the system.It has the advantages of simple structure,easy development and strong adaptability.The Model Free Adaptive Predictive Control(MFAPC)can further enhance the robustness of the system.The automatic overtaking scheme includes two parts: overtaking path planning and overtaking path tracking control.According to the information of itself and surrounding environment collected by on-board sensors,the unmanned vehicle plans a safe and feasible overtaking trajectory online,and then tracks the planned trajectory through the path tracking controller.Firstly,through simulation comparison,the path planning method based on quintic polynomial is determined as the overtaking path planning method in this thesis.Then,based on MFAC and MFAPC,two kinds of data-driven automatic overtaking control schemes on expressway are proposed.Finally,a Simulink/Car Sim co-simulation environment is built to verify the effectiveness of the two control schemes proposed in this thesis.The main work is as follows:(1)Automatic overtaking path planning method: Based on the 2-DOF vehicle kinematics model,the overtaking process is analyzed,the characteristics of quintic polynomial lane change path planning algorithm,trigonometric function path planning algorithm and circular arc lane change path planning algorithm are studied,and the quintic polynomial lane change path planning algorithm is determined through simulation comparison.(2)An overtaking path tracking control method based on MFAC: The overtaking system of unmanned vehicle is equivalent transformed into a full-format dynamic linearized data model,and the MFAC overtaking path tracking control scheme is proposed by using the data model,which mainly includes model-free adaptive control algorithm,pseudo-gradient estimation algorithm and reset algorithm.(3)Aiming at the problems of time delay and external interference in the unmanned vehicle overtaking system,combining the advantages of predictive control and model-free adaptive control,an overtaking path tracking control method based on MFAPC was designed.(4)Using the I/O data of the unmanned vehicle overtaking system,the traditional PID path tracking control method and the two proposed path tracking control methods are simulated and compared in the Simulink/Car Sim co-simulation environment.The results show that both MFAC and MFAPC algorithms can complete the tracking control of overtaking path.Compared with PID algorithm,MFAC and MFAPC algorithms have stronger tracking ability,higher tracking accuracy,and better ride comfort.