Research On The Longitudinal And Lateral Collaborative Control Codel Of CAV Fleet Based On GANs And Imitation Learning

In recent years,with the rapid development of connected and autonomous vehicles(CAV),CAV began to penetrate into the flow of traditional vehicles(MV),and gradually evolved into a mixed traffic flow.The control model of CAV has become a research hotspot in the field of intelligent transportation and intelligent vehicles.How to meet the safety,real-time and team stability of CAV under the driving behaviors of lane change and car following becomes the difficulty of CAV control model.Aiming at the problem of insufficient security and real-time in the longitudinal control of CAV,this paper puts forward the GANs longitudinal vehicle following control model,which is resistant to neural networks(GVFM).In order to improve the stability of CAV team in lateral control,the paper puts forward the GANs simulation learning lateral control model(GAIL-LCM).Finally,combined with the control strategy of CAV team,it puts forward the vertical / horizontal cooperative control of CAV team based on GVFM and GAIL-LCM.The main research work of this paper is as follows:(1)In view of the slow response of CAV to the change of front and rear vehicle speed in the longitudinal control process,which leads to the lack of safety,combined with the algorithm idea of GANs zero sum game,a GVFM composed of generator model(GM)and discriminator model(DM)is constructed,and its performance is verified by setting different permeability.The experimental results show that the accuracy of GVFM on test set is 95.3%,which can meet the requirements of safety,real-time and fleet stability better than other models;(2)In view of the current situation that the lateral control model of CAV deals with the slow driving which leads to the poor control stability of CAV fleet,the GAIL-LCM model is designed by using the excitation function and the methods of GANs and imitation learning.The results show that the accuracy of GAIL-LCM is 96.1%,97.2% and 98.6% for steering wheel angle,lateral acceleration and longitudinal acceleration on the test set,which effectively improves the problem of insufficient lateral control stability of CAV fleet;(3)In view of the poor stability of the vertical and horizontal collaborative control model of CAV fleet in the mixed traffic flow scenario,which is greatly affected by the surrounding vehicles,combined with the control strategy of CAV fleet,the vertical and horizontal collaborative control model of CAV fleet based on GVFM and GAIL-LCM is proposed,and compared with other control methods in multiple scenarios.The results show that under the cooperative control of GVFM and GAIL-LCM,the stability of motorcade is better,and the speed loss of motorcade is effectively reduced.In conclusion,this paper focuses on the improvement of road capacity in the mixed traffic scenario,and studies and verifies the longitudinal / lateral collaborative control of CAV fleet based on GVFM and GAIL-LCM on the basis of the research on the longitudinal / lateral optimal control of CAV.The research results further enrich the traffic flow theory and methods.