Lane-change Trajectory Planning Method For Driverless Vehicles In Multi-Vehicle Environment

Lane changing is a common and dangerous condition in the process of vehicle driving.Especially in the multi-vehicle environment,the relative safety between lane changing vehicles and surrounding vehicles is very important.In order to improve the safety and success rate on lane changing of unmanned vehicles in multi-vehicle environment,the vehicle trajectory prediction method and lane changing planning method of unmanned vehicles in multi-vehicle environment are studied in this paper.Firstly,the trajectory prediction model of the surrounding vehicles in multi-vehicle environment is studied.According to the existing trajectory prediction models based on Long Short-Term Memory(LSTM),a trajectory prediction model considering external parameters such as vehicle type and vehicle profile is proposed to improve the accuracy of trajectory prediction for vehicles under different conditions.In order to reduce the calculation amount of model prediction and improve the real-time performance,a dynamic sensitive area is proposed to filter the surrounding vehicles with low correlation with the target vehicles.The proposed model is trained on the NGSIM(Next Generation Simulation)dataset and compared with the Conv Social LSTM(Conv Social LSTM).The results show that the proposed model has a better performance in accuracy and real-time performance.Secondly,on the basis of the trajectory prediction,the lane-changing trajectory planning method in multi-vehicle environment is studied.According to the motion state information of self-vehicle,lane change trajectory cluster is obtained by using the quintic polynomial programming method,and the stable boundary of trajectory cluster is obtained according to the3-DOF dynamics model of self-vehicle.In order to obtain a safe lane-changing track in the track cluster,on the basis of the predicted track of the surrounding vehicles,the security objective optimization function is established separately for the surrounding vehicles to solve the optimal track in the track cluster.A simple lane-changing scenario is set to verify the rationality of the proposed lane-changing trajectory planning method.Finally,the reliability of the proposed lane-changing trajectory planning method in multivehicle environment is verified.On the basis of the vehicle 3-DOF dynamics model,a vehicle lateral controller based on model predictive control is established,and a simulation platform is built on Carsim and Simulink,and the effectiveness of the simulation platform is verified by a group of typical lane-changing data.In a multi-vehicle environment,lane-changing scenarios are set up in which the vehicle in front and the vehicle behind on the side interfere with the lane changing of the vehicle.The predicted trajectories and the real trajectories are respectively used as references to carry on lane-changing trajectory planning.The safety of the proposed lanechanging planning method is proved by analyzing the longitudinal and absolute distances between the vehicle and the surrounding vehicles in the process of lane changing.Through the analysis of vehicle lateral acceleration and front wheel rotation history in lane-changing process,it is proved that the proposed lane-changing planning method has a better lane-changing efficiency and comfort.