Research On Vehicle Steering Control Based On Visual Information And GIS

The improvement of vehicle intelligence can not only alleviate driver fatigue,but also effectively reduce human-induced traffic accidents.Therefore,the study of environmental perception and intelligent control technology has an important impact on improving people's travel.Based on visual inspection technology and GIS technology,this topic obtains external environmental information such as lane lines,traffic lights,vehicles in front,and driving routes.This paper focuses on the vehicle steering control problem,and studies the lateral and longitudinal motion control of vehicles.Based on camera calibration technology and Hough tra nsform principle,the vehicle heading angle calculation method,lane line and parking line detection algorithm are designed.The vehicle heading angle,lane type and distance from parking line are obtained,which are used for vehicle heading control,vehic le positioning in lane and vehicle positioning at intersection.A signal lamp detection method based on color and feature is designed to detect circular and arrow signals quickly.A vehicle recognition and ranging method based on the combination of vehicle bottom shadow and license plate location is designed to realize vehicle longitudinal control.Based on GIS technology,a road traffic map database and a GIS map are established to store road and signal attributes and spatial location information for path planning algorithm and intersection navigation control.An improved Dijkstra path planning algorithm is designed to calculate the vehicle path quickly and guide the vehicle.The vehicle location algorithm in multi-lane based on virtual-real lane detection algorithm,GPS positioning technology and GIS map technology is designed to realize the vehicle positioning function in multi-lane.Based on GIS map technology and signal detection technology,a signal detection strategy is designed.A method of determining the region of interest by look-up table is also designed.An intersection location and navigation algorithm based on vision detection technology and extended Kalman filter algorithm is designed to control vehicles passing through the intersection.Based on the model predictive control algorithm,a vehicle lateral motion and longitudinal motion controller is designed.Lateral displacement deviation,heading angle deviation and its change rate are selected as state variables,and vehicle steering angle as control variables to realize lateral motion control of vehicles.Longitudinal distance deviation,velocity deviation and vehicle acceleration are selected as state variables and expected acceleration as control variables to realize vehicle longitudinal motion control.In addition,vehicle longitudinal control strategy is designed to follow driving conditions and intersection conditions.Using MATLAB / Simulink and CarSim to establish a co-simulation platform,the lateral motion controller and the longitudinal motion controller are simulated respectively.The results show that the controller has good accuracy and real-time.The real vehicle test results also show that the controller has good control performance.