Research On Lane-changing Obstacle Avoidance Control Based On Driving Behavior Recognition Of Preceding Vehicle In Adjacent Lanes

The traffic environment is a complex system that multiple traffic participants influence each other and change dynamically.On the expressway,the velocity of vehicle is fast,the conflicts between vehicles are frequent and the traffic accidents are harmful.The dangerous lane-changing or overtaking action of the preceding vehicle in adjacent lanes will affect the normal driving of the intelligent vehicle.This paper proposes a lateral lane-changing obstacle avoidance control technology based on driving behavior recognition of preceding vehicle in adjacent lanes.This technology can recognize the cutting-in behavior of the preceding vehicle in adjacent lanes in advance,and implement obstacle avoidance control for the dangerous cutting-in behavior of the preceding vehicle in adjacent lanes.Firstly,aiming at the problem of recognizing the cutting-in behavior of the preceding vehicle in adjacent lanes,a driving behavior recognition method based on Gaussian Mixture-Hidden Markov Model(GM-HMM)is proposed.Based on the characteristics of vehicle lane-changing and driver’s decision-making in expressway scene,the longitudinal velocity,the lateral displacement in 0.5 second and the lateral velocity of preceding vehicle in adjacent lanes are selected as observation variables,the Baum-Welch algorithm and forward-backward algorithm are applied to model and recognize three driving behaviors(including lane keeping,left lane-changing and right lane-changing),and the recognition accuracy of three driving behaviors in different recognition time windows is tested.The test results show that GM-HMM can quickly and accurately recognize the cutting-in behavior of the preceding vehicle in adjacent lanes,and can provide sufficient information and time for subsequent driving environment safety assessment and intelligent vehicle decision-making.Then,aiming at the safety assessment problem of cutting-in threat vehicle from adjacent lanes,a braking safety distance model based on the driving intention parameters of threat vehicle is proposed.When the lane-changing cutting-in behavior is judged as danger,the intelligent vehicle will choose the appropriate obstacle avoidance method according to the driving environment.Therefore,the minimum lane-changing safety distance model is established to judge the safety of lane-changing of intelligent vehicle.The intelligent vehicle can choose lateral lane-changing obstacle avoidance when the front and rear vehicle on the target lane meet the requirements of the minimum lane-changing safety distance.And then,aiming at the dangerous cutting-in behavior of preceding vehicle in adjacent lanes,a lateral lane-changing obstacle avoidance controller is designed based on constrained control algorithm.Firstly,the vehicle motion is reasonably simplified and the dynamic equation of the two degree of freedom vehicle model is builded.Secondly,the ideal front wheel steering angle is calculated according to the designed lateral acceleration,and based on the asymmetric Barrier Lyapunov Function(BLF),combine backstepping and dynamic surface control technology to design the vehicle direct yaw moment controller to better track the desired trajectory.The simulation results on MATLAB/Simulink show that the designed controller can stably and safely change lanes to avoid obstacles.Finally,the experiment section introduces the process of sample data collection,shows the modeling training code and model test code of driving behavior recognition model based on MATLAB,and tests the effectiveness of lateral lane-changing obstacle avoidance controller.