关键词: 排爆机器人, 双目立体视觉, 手眼系统, 摄像机标定, 特征匹配, 三维重建

Abstract:

The existing explosive ordnance disposal （EOD） robots are usually operated manually and are hard to operate. In order to solve this problem, this paper develops a single-hand and binocular eye-in-hand system based on binocular stereo vision for the semi-intelligent EOD robot. The system uses Matlab7.0 as the calculation engine and adopts the pattern matching library EasyMatch in machine vision software eVision6.2 to perform the stereo matching. It can capture real-time images and make camera calibrations, image preprocessing as well as feature point matching. It can also determine the coordinate of suspicious objects, display real-time images on the console, and control the semi-intelligent EOD robot to grasp suspicious objects automatically and accurately. Experimental results show that the proposed system is of a relative error of less than 2.54% , which meets the practical requirements of EOD robots.

Key words: explosive ordnance disposal robot, binocular stereo vision, eyes-in-hand system, camera calibration, feature matching, 3D reconstruction