Integrated Lane Change Control Scheme For A Heterogeneous Platoon Of Connected And Autonomous Vehicles

As an important application,platoons of Connected and Autonomous Vehicles(CAVs)have great advantages in reducing traffic congestion,improving driving safety and enhancing fuel economy.Platoon need to change lanes for its destination or changes in road structure.Lane change strategy,safe lane change path planning and vehicle lane change control are crucial elements in lane change behavior.Compared with single vehicle lane change,platoon needs to consider its internal safety,it is necessary to maintain a proper spacing during lane change.safely path planning and platoon lane change control methods need to be satisfied.The lane change of heterogeneous CAVs is taken as the research object in this paper.Lane change path planning,Distributed model predictive longitudinal control based on nonlinear dynamic model,vehicle longitudinal control based on reinforcement learning and general lateral model predictive control based on vehicle kinematics are mainly studied.Based on the accurate modeling and analysis of the vehicle braking process,the safety distance under different velocity is obtained,then a high-order Bessel lane change curve control-points selection method with the lateral acceleration constraint is proposed and an optimal curve selection method is designed.In actual scene,platoon hardly meets the ideal situation that vehicle type and dynamic parameters are exactly the same and known.In order to reach accurate longitudinal control,nonlinear dynamics modeling is carried out for small cars which are easy to obtain their dynamic models and parameters,then the distributed model predictive controller is designed for small vehicle controlling.For heavy duty vehicles such as trucks which are difficult to obtain dynamic model parameters and whose load mass and load distribution often change,a DDPGbased reinforcement learning method is designed to learn the control strategy.The simulation of the longitudinal and lateral controller of heterogeneous platoon of CAVs is designed.The numerical simulations under homogeneous and heterogeneous platoons are carried out using the longitudinal controller,with random initial vehicle distance and speed as well as the fluctuate leader vehicle speed profile.A comparative test is carried out on the designed lateral controller,proved that the controller has better control accuracy among similar controllers.In terms of longitudinal control based on reinforcement learning,Double DQN and DDPG are used,the corresponding network structure,state space,action space and reward function are designed respectively.The simulation results show that the DDPG-based strategy is capable of achieving expected platoon control.