Target Measurement Method Based on Sparse Disparity for Live Power Lines Maintaining Robot

3D perception has been an indispensable part of the research on many manipulator robots. In this paper, we propose a the new kind of target measurement method to perceive the surrounding environment of robots based on binocular stereo vision theory, which can effectively improves the 3D reconstruction accuracy and effectiveness of the target recognition. The sparse disparity method we apply is different from the dense binocular stereo vision method in common. At the beginning of the article, a series of effective methods of image processing we use to identify and extract target images will be introduce. A novel operator-matching algorithm is proposed to improve the extraction and recognition. The contour-center stereo matching theory is expounded to lay a foundation to the calculation of sparse disparity estimation. To give accurate 3D reconstruction model information, a calibration system with low measurement error is proposed based on dual manipulators of our robot. After coordinate transformation implementation, robots obtain point cloud data of the target and able to perform interactive actions. A wiring experiment is set to test this method by judging work efficiency and success rate.