Visual Servoing Invariant to Changes in Camera-Intrinsic Parameters

Abstract

This paper presents a new visual servoing scheme which is invariant to changes in camera-intrinsic parameters. Current visual servoing techniques are based on the learning of a reference image with the same camera used during the servoing. With the new method, it is possible to position a camera (with eventually varying intrinsic parameters), with respect to a nonplanar object, given a "reference image" taken with a completely different camera. The necessary and sufficient conditions for the local asymptotic stability show that the control law is robust in the presence of large calibration errors. Local stability implies that the system can accurately track a path in the invariant space. The path can be chosen such that the camera follows a straight line in the Cartesian space. Simple sufficient conditions are given in order to keep the tracking error bounded. This promising approach has been successfully tested with an eye-in-hand robotic system.