Research On Obstacle Detection And Automatic Emergency Braking Strategy On Slippery Road

The intelligent vehicles driving system integrates multiple technologies such as intelligent transportation,communication technology,and computer technology.Intelligent driving technology provides important technical support for the safe operation of intelligent vehicles.In order to reduce traffic congestion and traffic accident rate in a wide range,the environment perception technology of intelligent vehicles has become a key research direction,and a variety of algorithms for detecting obstacles in front of general roads have been applied and achieved good results.However,under the deteriorating driving environment such as slippery and reflective roads,the detection accuracy of obstacle detection methods will be greatly affected.This paper had used a vision camera to obtained environmental information real-time.This paper had used Bayesian classifier to train the texture and color features of dry and slippery road conditions,so as to realize the classification of road conditions.In order to output the position information of all obstacles in the reflective road environment accurately,this paper had proposed an improved VM obstacle detection model(combined detection of improved VIDAR and machine learning)to realize the process of accurately detecting obstacles.In this paper,the Kalman filter tracking algorithm had been used to complete the position prediction of the moving target,and the safety state assessment had been carried out by identifying and predicting information to select a braking strategy.Based on the above introduction,the main research contents of the paper are as follows:Firstly,in order to improve the accuracy of obstacle detection,an improved VIDAR obstacle detection method had been proposed.In order to improve the speed of obstacle detection,this paper had combined the object recognition algorithm of the deep learning Faster R-CNN to jointly detect road obstacles.Due to the false obstacles that may exist in the reflective environment of slippery roads,this paper had built an obstacle detection model called improved VIDAR.(1)Confirm stereo unknown obstacles.Machine learning had been used to identify known obstacles,then obtained images after removing the known obstacles,and used VIDAR to screen out stereo unknown obstacles.(2)Construct obstacle rectangle.Found the feature points on the obstacle that is the farthest in the horizontal and vertical directions to construct a rectangle.(3)Calculate the horizontal distance.Calculated the horizontal distance from the object point to the camera according to VIDAR.(4)Identify stereo obstacles.calculated the width of obstacle rectangle,and determined the relationship between the height and width through the triangle similarity principle,calculated the rectangular height of real and pseudo obstacles by using the same width when vehicles and obstacles moving.Calculated and compared the height values,and identified stereo obstacles.Secondly,based on the information of obstacle detection,a two-stage automatic braking strategy(including full braking and partial braking)had been designed.By analyzing the safety distance model of full braking and the safety distance model of partial braking and comparing with the distance information of the detected obstacles,the braking type had been selected.When determined the safety distance model of full braking,we had considered it as the minimum safety distance of ego-vehicle.When determined the safe distance model of partial braking,we had determined the most comfortable braking deceleration curve of the driver through experiments,and obtained the safe distance through the quadratic integration of the acceleration curve.Thirdly,the feasibility of two-stage automatic braking strategy had verified by simulation and experiment.The model of the brake motor had been determined and selected,and the real-time torque of the motor had determined.The values of the brake fluid pressure had been obtained by MATLAB/Simulink.In the experimental,the real-time torque of the motor had been determined by the braking safety distance and the distance of the obstacle.The torque information had been transmitted to the STM32,so that it had produced the corresponding signal to control the motor torque,and finally realized the braking effect.The actual braking effect has been determined by collecting the hydraulic pressure of the main cylinder,and the effectiveness of the two-stage automatic braking strategy has been verified by comparing with the simulation effect.