Discrete-time robot visual feedback in 3D positioning tasks with depth adaptation

Abstract

An adaptive visual feedback scheme is designed to perform 3D positioning tasks. The dynamic camera-object interaction model is derived in discrete time, since the visual sampling time is not negligible at the current state of technology. Active contours are used to track the 2D projection of the visible object's surface in the image plane. Uniform asymptotic stability of the image reference set-point is proved using the Lyapunov direct method, and a 3D estimation procedure, based on prediction errors, is used to cope with the unknown depth of the object. Experimental results with a 6-DOF robot manipulator in eye-in-hand configuration validate the theoretical framework in real conditions.