Lane Change Trajectory Planning And Tracking Control Based On Driving Style

Local trajectory planning and trajectory tracking control are of great significance to the realization of autonomous driving technology.At present,the characteristics of individual drivers are seldom considered when designing autonomous driving systems.This paper combines the characteristics of different driving styles of lane changing with local path planning to meet the needs of different drivers’ ride experience sense while ensuring the safe driving of vehicles.The main research work includes:(1)Recognition model of driving styles considering different traffic densities and driving behaviors.Starting from the influence of traffic density and driving behavior on driving style recognition,a simulated driving simulation experiment platform was built using Prescan,Matlab/Simulink and NGSIM datasets to collect driving data of drivers in straight lanes with different traffic densities and to classify driving behaviors and extract feature parameters of corresponding driving behaviors.Principal Component Analysis(PCA)was used to post-process the feature parameters,and K-means++ was used to cluster the different driving behaviors under different traffic densities.Based on the results of the clustering,a driving style recognition model was established using the random forest algorithm and compared with the driving style recognition model without segmented traffic density and driving behavior.The results show that the driving style recognition model can accurately identify the comprehensive driving style of drivers and can adapt to different traffic environments and driving behaviors.(2)Lane change trajectory planning based on Dynamic Programming(DP)and Quadratic Programming(QP)algorithms and considering driving styles.In the Frenet coordinate system,the DP and QP algorithms are used to generate a safe and smooth path and speed profile respectively by combining the primary vehicle and obstacle information and drawing SL and ST maps based on the reference path.The lane change characteristics of different driving styles are added to the QP algorithm,and the transverse lane change distance of different driving styles is used as a combination point to represent the cost function of lane change characteristics of different driving styles,and the lane change longitudinal distance of different driving styles is used to judge the lane change timing to reasonably assign the weight coefficients.(3)The trajectory tracking control based on feedforward Linear Quadratic Regulator(LQR)with double discrete proportional integral differential control(PID)algorithm.For the lateral control,the differential equations of motion of the two-degree-of-freedom vehicle model are derived first,and then the differential equations of the state error in the discrete state of the path are derived using the path tracking error module;the controller gain and the front wheel turning angle and the front wheel turning angle increment are calculated using the feedforward LQR controller module.For the longitudinal control,a double discrete PID tracking control algorithm is used to make the longitudinal distance error as well as the speed error converge to zero and obtain the corresponding throttle opening and brake pedal pressure.(4)Verification and analysis of algorithm effectiveness based on joint simulation platform.Prescan,Carsim and Matlab/Simulink are used to build an integrated simulation platform for perceptual decision planning and control.Firstly,the safety effectiveness of the lane change trajectory planning and trajectory tracking control algorithm without considering the driving style is verified,and the improved trajectory planning algorithm is also verified based on the traffic scenario with 70% traffic density that has been built,establishing the scenario of avoiding static and dynamic obstacles,and combining the data of lane change characteristics of different driving styles in this scenario.The results show that intelligent vehicles can achieve safe,smooth and personalized lane changing and obstacle avoidance.