Research On Binocular Visual Odometer Localization Algorithm For Tunnel Detection

The tunnel detection robot replaces the operator to enter the tunnel to detect safety hazards.The binocular vision localization algorithm can locate the specific location of the safety hazards,which serves as one of the important methods of robot positioning.Therefore,it is of great practical significance to realize the precise location of the tunnel detection robot in the tunnel environment through the binocular vision localization algorithm.This paper mainly studies the binocular vision localization algorithm based on feature point matching.The main contents of the research are as follows:(1)Firstly,establishing a monocular camera pinhole model,based on which a binocular camera pinhole model is established.According to Zhang Zhengyou calibration method to complete the stereo calibration of the binocular camera to obtain the internal and external parameters of the camera,and then complete the stereo correction of the binocular camera according to the internal and external parameters,so that the left and right images of the camera reach the coplanar alignment,and finally the camera calibration error analyzed.(2)After graying preprocessing of the image,the SIFT,SURF and ORB feature point extraction algorithms are analyzed.The image feature points are then extracted by the ORB operator,matched by the stereo matching constraint for the left and right images in the same frame,tracked by the motion model for the images in the two frames before and after,accelerated by the word bag model when the motion model fails.After the matching is completed,the mismatching point is eliminated by the RANSAC algorithm to improve the positioning accuracy.The motion estimation of the camera is performed by the ICP(Iterative Nearest Point)algorithm,and the ICP problem is solved by the SVD algorithm to obtain the rotation matrix and the translation vector of the latter frame relative to the previous frame,thereby realizing the positioning of the robot.At the same time,due to the motion estimation error,the motion trajectory is optimized by the graph optimization algorithm.(3)Perform visual odometer positioning experiment in the construction of 11 m simulated tunnel environment and 50.4m indoor corridor environment.During the advancement of the machine,the actual coordinates and the coordinates estimated by the algorithm are recorded according to the fixed separation distance.The root mean square error(RMSE)is used to judge the accuracy of the Positioning.The results show that the binocular vision localization algorithm can be used for the location of the tunnel detection.