Research On Avoidance Obstacle Strategy Of Coal Mine Underground Rescue Robot Based On Binocular Vision

In this thesis,aiming at the problem of avoiding obstacle in coal mine underground rescue robot,a autonomous obstacle avoidance scheme is put forward,which combines the three-dimensional physical information of the obstacle recovered by the binocular vision and the fuzzy control algorithm to avoidance the obstacle.Firstly,the traditional SIFT matching algorithm is improved,and the superiority of the improved algorithm in matching accuracy and matching efficiency is verified through simulation experiments.Then the triangulation method is used to perform three-dimensional reconstruction of obstacles,and finally the obstacles are based on binocular vision extraction.A three-dimensional physical information system is designed and a set of obstacle avoidance schemes based on fuzzy control theory is designed.Through simulation experiments and experimental tests,the results show that the binocular vision technology designed in this thesis combined with the fuzzy control obstacle avoidance scheme can safely and effectively make the rescue robot reach the target location.The following are the main research contents of this thesis:(1)This thesis have more introducing for the principle of binocular camera imaging,binocular distance measurement,camera calibration and other principles.The principle of Zhang Zhengyou camera calibration method is studied in detail,and using MATLAB simulation software,the camera is calibrated and the camera's internal and external parameters are obtained.(2)This thesis focuses on the SIFT matching algorithm and the improved SIFT matching algorithm and experimental analysis to have more introduction.Firstly,the SIFT operator is simplified,and the SIFT feature vector is reduced from the original 128-dimensional to 24-dimensional,which reduce the operation time of the algorithm.At the same time,combining with the the principle of epipolar geometry constraint to reduce the range of matching search from two-dimensional to one-dimensional,so that the matching search time is shorteneda After the first matching,combining with the RANSAC algorithm to eliminate false matching points,to further improve the matching accuracy.(3)Using the principle of three-dimensional reconstruction of binocular vision,the three-dimensional reconstruction of the feature points of obstacles is performed,and three-dimensional physical information such as distance,height,and outline of obstacles is extracted.Based on the extracted three-dimensional physical information of obstacles,a set of obstacle avoidance solutions based on fuzzy control theory is designed.The MATLAB simulation results show that the fuzzy obstacle avoidance system can safely and effectively make the robot reach the target position.Finally,using the idea of modular programming to have a experiment on the experimental platform of the binocular vision robot,the stereo vision technology and the fuzzy obstacle avoidance scheme mentioned in this thesis are respectively verified by experiments.The results show that the obstacle avoidance scheme studied in this thesis which combine with the stereo technology and the fuzzy control has a good result on validity and practicability.