Research On Local Lane Change Decision And Lane Change Trajectory Planning Algorithm Of Driverless Vehicle

As the direction of modern automobile development,intelligent driving vehicles have great advantages in improving road utilization,driving efficiency,safety,comfort and other aspects.However,with the increase of car ownership,traffic congestion,traffic accidents and other problems have become prominent.Safe,efficient and reliable intelligent driving vehicles have become an urgent need for social development.In the dynamic traffic flow scenario,how to make accurate and effective lane change decisions and plan a reasonable lane change track based on the driving information of the surrounding vehicles is the key technology to realize intelligent driving.Therefore,it is of great significance to study lane change behavior decision and lane change trajectory planning of intelligent driving vehicles.In this paper,the driverless vehicle is taken as the research object,and the lane change behavior decision and lane change trajectory planning algorithm in dynamic complex traffic flow scenarios are mainly studied,and simulation experiments are designed to verify and evaluate the algorithm.The main research contents of this paper are as follows:First,the constraint conditions of the vehicle during driving are studied.By analyzing the state of the two-degree-of-freedom vehicle dynamics model in the process of driving,and combining the relevant vehicle theoretical knowledge and geometric relations,the dynamic constraints of vehicle trajectory planning are established.Constrain the geometric relationship between vehicle driving path and road boundary,and analyze the safety of vehicle driving.These provide necessary constraints for designing local lane change trajectory planning algorithm.Then,the lane change behavior decision of vehicles in dynamic traffic flow scenario is studied.This paper analyzes the driver’s lane change process,and divides the lane change process into three stages: the generation of lane change intention,the execution of lane change action and the end of lane change,and determines the time required for the generation of lane change intention and the execution of lane change action.In the three-lane driving environment,the characteristic factors that affect the decision of lane change behavior are analyzed and extracted,and then the classification rules are established to preprocess the vehicle lane change behavior.In order to improve the effectiveness of lane change behavior decision,the vehicle lane change behavior is analyzed and the lane change benefit model is established.The excellent lane change behavior decision is selected through velocity space benefit,distance space benefit and risk perception benefit.The BP neural network model is built to learn and train the effective lane change decision behavior to obtain decision maker.The results show that the neural network has high performance accuracy and can choose a more appropriate time for lane change.In the dynamic traffic scene on the structured road,the trajectory planning of local lane change is carried out after obtaining the lane change behavior decision.Firstly,according to the driving status of surrounding vehicles in the dynamic traffic scene,determine the selected area of the lane change target point.Then,taking into account the comfort,safety and smoothness of the lane change track,quintic polynomial track changing trajectory planning algorithm based on the multi-objective value function is designed,which transforms the problem of selecting the lane change target point into the problem of seeking the optimal solution of the five-degree polynomial planning track value.The research results show that the quintic polynomial lane change trajectory planning algorithm based on multi-objective value function can obtain better smoothness and comfort of lane change trajectory under different working conditions.Finally,different traffic flow scenarios are designed,and the proposed lane change behavior decision algorithm and lane change trajectory planning algorithm are simulated and analyzed.The simulation platform is built by Prescan and Matlab/Simulink,and the dynamic traffic flow scene and vehicle model can be built in the Prescan.Then,the lane change behavior decision and lane change trajectory planning algorithm are written in Matlab/Simulink.The algorithm is evaluated and analyzed by the comfort,minimum safety distance and curvature of the lane change trajectory.The results show that the BP neural network based on the lane change revenue model can select better lane change time points,and the lane change trajectory based on the quintic polynomial programming of the multiobjective value function has better performance in terms of safety,smoothness and comfort.