Control of the Final-Phase of Closed-Loop Visual Grasping using\n Image-Based Visual Servoing

Abstract

This paper considers the final approach phase of visual-closed-loop grasping\nwhere the RGB-D camera is no longer able to provide valid depth information.\nMany current robotic grasping controllers are not closed-loop and therefore\nfail for moving objects. Closed-loop grasp controllers based on RGB-D imagery\ncan track a moving object, but fail when the sensor's minimum object distance\nis violated just before grasping. To overcome this we propose the use of\nimage-based visual servoing (IBVS) to guide the robot to the object-relative\ngrasp pose using camera RGB information. IBVS robustly moves the camera to a\ngoal pose defined implicitly in terms of an image-plane feature configuration.\nIn this work, the goal image feature coordinates are predicted from RGB-D data\nto enable RGB-only tracking once depth data becomes unavailable -- this enables\nmore reliable grasping of previously unseen moving objects. Experimental\nresults are provided.\n