Research On Lane Departure Warning And Lane Keeping Algorithm For Vehicle

With the continuous deepening of the research on the intelligentization in automotive industry,the development of advanced driver assistance systems has become more and more important.The Lane Departure Warning system(LDWS)and Lane Keeping Assist system(LKAS)are two basic types of advanced driver assistance systems,LDWS and LKAS can effectively reduce accidents caused by lane departures when the driver is tired or inattentive.In view of complex and changing driving environment,the LDWS and LKAS should have the ability to accurately identify lane information under disturbance information and timely send deviation warning to the driver.The system should keep good ability to active steering control under the low,middle and high speed conditions.Based on the parameters of Tesla Model S models,this paper studies lane departure warning and lane keeping algorithms.The lane detection based on image information is studied in this paper.The system takes the image information of roads around campus as input.In the MATLAB,the image preprocessing,segmentation,edge detection and other operations are completed.Finally,a parameterized lane model can be obtained.This paper also analyzes Calibration algorithm of the camera,map-road coordinate system transformation and establishes a lane centerline deviation model based on image information.The FOD lane departure warning detection algorithm based on the preview offset is adopted.FOD can adjusts the preview time and the preview position offset according to different driving habits and tendencies of drivers.The simulation conditions are designed according to the national standard and implemented on the CARSIM platform.Finally,the lane departure warning judgment algorithm designed was verified.Using the principle of model predictive control and taking the requirements for the vehicle's running stability during lane keeping tracking into account,this paper studies the predictive control process with multiple sides of centroid slip angle,tire side slip angle,yaw rate and front wheel angle.A multi-constrained lane keeping model predictive controller is designed.According to the working characteristics of lane keeping active control process,we design rapid initializing condition with initial offset and double tracking trajectory tracking condition.The lane maintenance work stage corresponding to the two working conditions is selected.The model predictive controllers with different parameter conditions are designed.Combining the simulation results at low,middle and high speeds,we analyze the controller's rapid centering ability and the trajectory tracking effect.It is considered that the designed controller can quickly realize lane centering and achieve Lane trajectory tracking on full-speed conditions.The effect of different predictive time domain and control time domain on the trajectory tracking effect is analyzed.It is considered that increasing predictive time domain or reducing control time domain can decrease the control sensitivity and improve the control stability.The reasons for overshoot and oscillation in high-speed conditional trajectory tracking are analyzed.It is considered that the larger predictive time domain should be used to improve the stability of the control under high-speed conditions,and the smaller predictive time domain should be used to reduce the computational burden under low-speed conditions.Based on this idea,a predictable time-domain model predictive controller was designed and joint simulated by SIMULINK and CARSIM.The simulation results show that the model predictable controller with predictable time-domain and multi-constraint can keep good stability of vehicle at different speeds during the lane keeping track process.