Research On Automatic Driving Vehicle Trajectory Planning Algorithm In Lane Change Scene

With the rapid increase in the number of automobiles,cars have become the preferred means of transportation for adults.However,along with the progress of traffic activities,a large number of traffic accidents have also occurred.As an important component of Intelligent Transportation Systems(ITS),autonomous vehicles play a significant role in improving travel efficiency and reducing traffic accidents.Lane changing scenarios have always been a focus in autonomous driving technology.During the driving process of autonomous vehicles,this paper focuses on the rational selection of lane changing timing and the generation of safe and comfortable lane changing trajectories based on dynamic road and surrounding environmental conditions.First,research was conducted on lane-changing decisions and lane-changing areas,with the research object set as the problem of free lane-changing for autonomous vehicles during driving.The concept of cumulative driving dissatisfaction and optimal vehicle spacing is proposed,along with the calculation method.Cumulative driving dissatisfaction is used to describe the degree to which the speed and distance of autonomous vehicles during driving in the original lane fail to meet the expected target,while the optimal vehicle spacing provides a basis for selecting the target lane.Based on this,the lane-changing process is divided into a straight-line stage and a merging stage,and by sampling the acceleration and merging time within a certain range,the merging start point and merging end point are calculated,providing reference speed and lane boundary constraints for the lane-changing process.Based on the comparison of various lane-changing trajectories,a segmented fifth-order polynomial curve is used as a component of the reference trajectory.At the beginning of the lane-changing process,a lane-changing reference trajectory is generated based on the "sampling→evaluation→search" approach.This method samples the lane-changing area in the Frenet coordinate system,connects the sampling points using a fifth-order polynomial curve,evaluates the trajectory curve given the known longitudinal reference speed,and obtains the reference trajectory through dynamic programming search.This paper transforms the lane-changing trajectory optimization problem into a constrained nonlinear optimal control problem.The iterative linear quadratic regulator(ILQR)algorithm is used to solve the optimization control problem by eliminating constraints through obstacle functions.Control limits,dynamic obstacles,and lane boundaries are considered in the problem-solving process to address their impact on the lane-changing trajectory.To deal with the constraints imposed by complex lane boundaries,a new constraint method is proposed in this paper,which uses the tangent of the projection point of the trajectory on the lane boundary instead of the local lane boundary at that point,simplifying the form of the lane boundary constraint..The trajectory planning simulator TPSim was used to validate the proposed method by simulating two lane-changing scenarios: background vehicles with constant acceleration and variable acceleration.Experimental results show that the proposed free lane-changing decision method can choose the appropriate time to change lanes based on traffic conditions,and the lane-changing trajectory planning can generate trajectories that meet the vehicle’s kinematic characteristics and provide a comfortable and safe ride.The trajectory can also be adjusted in real-time based on the dynamic environment..