Object Detection of Non-cooperative Target for Dual-arm Manipulation

Abstract

Aiming at the task of capturing the docking ring of non-cooperative target in space, a method for object detection of non-cooperative target based on dual-arm manipulation is proposed. In addition, this paper proposed a system based on visual servo using two arms to capture the docking ring. In this system, the construction of robots formed by two UR manipulators provides services for non-cooperative satellites and meets the basic requirements of space non cooperative target acquisition task. By using Eye-in-Hand cameras and feature-based object detection algorithm, the active measurement of uncooperative spacecraft is realized. The experimental results show that Eye-in-Hand cameras and visual algorithm can be used to extract and track the target features for the non-cooperative target without hook and mark. This active visual guidance method based on features is proved to be feasible.