Research On Lane-changing Trajectory Planning Of Intelligent Vehicles Under Multi-vehicle Interaction

With the widespread application of artificial intelligence and internet technology in the automotive industry,intelligent driving vehicles have entered our lives.How to safely and effectively achieve intelligent driving in complex environments has become a major challenge.Since the proposal of intelligent driving technology in the last century,it has remained a research hotspot among universities and enterprises,including autonomous lane change for vehicles.The intelligent driving autonomous lane change system includes four modules: obstacle perception,lane change decisionmaking,lane change trajectory planning,and tracking control.Among them,lane change decision-making and lane change trajectory planning are two key technologies in the research and development of intelligent driving vehicles.This thesis focuses on the lane change scenario of intelligent driving vehicles with multiple vehicle interactions,and designs an intelligent driving vehicle lane change decision-making algorithm and a lane change trajectory planning algorithm.The effectiveness of the algorithm is verified through joint simulation and real vehicle experiments.The specific research content of this thesis is as follows.(1)Design of lane change decision-making algorithm based on XGBoost.Firstly,the vehicle’s own driving state information and surrounding traffic vehicle’s driving state information are extracted from the NGSIM dataset.The Symmetric Exponential Moving Average(SEMA)filtering algorithm is used to filter the data,and the lanechanging vehicles and straight-driving vehicles are selected based on the change of lateral displacement and lane number within a certain time period.Then,the lane change decision-making variables are designed to determine effective variables.Finally,a lane change decision-making simulation experiment is established.The XGBoost lane change decision-making model is trained based on the effective variables,and the credibility of the lane change decision-making model designed in this thesis is verified according to the evaluation index.(2)Design of lane change trajectory planning algorithm based on cubic B-spline.This thesis selects commonly used lane change trajectory methods,establishes lane change safety constraints based on the rectangular vehicle model,and designs a lane change trajectory planning algorithm based on cubic B-spline algorithm according to the lane change instructions generated by the decision-making model.Then,based on the length of the lane change trajectory and the average curvature,the NSGA-Ⅱ multiobjective optimization algorithm is used to optimize the control point coordinates of the cubic B-spline curve,and the optimal lane change trajectory is obtained.Finally,the single-point preview tracking control model is used to track and control the lane change trajectory,and the effectiveness of the lane change trajectory is verified.(3)Simulation verification of lane change in multi-vehicle interaction scenarios.A simulation scenario of multi-vehicle interaction was established through PrescanSimulink-Carsim co-simulation to verify the lane change trajectory planning of intelligent driving vehicles.The results showed that the cubic B-spline trajectory planning algorithm optimized by NSGA-Ⅱ can effectively plan the lane change trajectory,and the single-point preview model can effectively track the reference trajectory.(4)Real vehicle experiments were conducted on a certain model of steer-by-wire chassis.The experimental platform,algorithm principles,and experimental procedures were described.The effectiveness of the three-degree B-spline trajectory planning algorithm based on NSGA-Ⅱ multi-objective optimization was verified using the pure pursuit algorithm in Autoware.AI software.