Research On Lane Keeping Assistance System Based On Model Predictive Control

With the advancement of automotive intelligence,advanced assisted driving systems have received extensive attention from researchers and many domestic and foreign automakers in recent years.As an important part of the advanced assisted driving system(ADAS),the lane keeping assistance system(LKAS)can actively control the vehicle to return to the original driving lane when the driver unconsciously manipulates the vehicle and causes the vehicle to deviate from the lane.The system is of great significance to reduce the burden on drivers,avoid traffic accidents caused by driver factors and improve vehicle safety.Based on this,this subject has studied the lane keeping assistance system,including the lane departure warning algorithm and lane keeping assistance control algorithm research as well as the lane keeping assistance control hardware-in-the-loop test bench.Firstly,the overall technical requirements of the lane keeping assistance system were studied,and the overall framework of the lane keeping assistance system based on the electric power steering system(EPS)was established;aiming at the problem of seamless switching between the two working modes of electric power steering system,the researches on power steering control strategy,active steering control strategy and the coordinated control strategy of these two modes were carried out.Considering the problem of man-machine conflict,a driver intention recognition module was designed.Secondly,after comprehensively analyzing the advantages and disadvantages of the currently studied early warning algorithms,this subject selected TLC for detailed study,considers different driver types,and set different virtual early warning boundaries in accordance with laws and regulations,and built a dynamic TLC model;through fuzzy inference to determine the danger of the vehicle deviating from the lane at the current moment,avoiding the phenomenon of excessively high false alarm rate of the lane departure warning system caused by a single TLC threshold.Thirdly,a lane keeping assistance control algorithm based on particle swarm optimization-model predictive control(PSO-MPC)was proposed.Based on the preview theory,combined with the linear two-degree-of-freedom vehicle model,the vehicle road model was built and used as the algorithm’s prediction model.The vehicle’s yaw angle,yaw rate,lateral speed,and lateral deviation at the preview distance l were used as the state quantity,and the steering wheel angle was used as the control quantity.The objective function was designed by considering the lateral deviation of the vehicle,the front wheel angle,the yaw rate of the vehicle,and the lateral offset of the center of mass.This objective function was used as the fitness function in the iterative optimization process of particle swarm optimization,and iteratively solved the fitness value of each particle to find the optimal particle(the best front wheel angle).A joint simulation model based on Matlab/Simulink,PreScan and CarSim was built,and the effectiveness of the algorithm was verified under simple straight and curved road conditions.Finally,a test bench for a lane keeping assistance system based on an electric power steering system and a camera-in-the-loop was built.The overall architecture of the test bench,as well as the calibration,debugging and testing of each hardware part were introduced in detail.After the hardware parts meet the accuracy requirements,the on-loop verification of the lane keeping assistance control algorithm was carried out,and the test verification was carried out under different working conditions with different speed ranges.The test results all reflected the effectiveness of the designed lane keeping assistance control algorithm.