Effect of camera calibration errors on visual servoing in robotics

Abstract

This paper addresses the problem of the influence of camera calibration errors on the performances of a control algorithm. More precisely, we examine the effects of errors in the values of intrinsic and extrinsic camera parameters on the stability and the transient behavior of a visual servoing scheme. After having recalled the basic principles of the visual servoing approach, we derive a general form for the interaction matrix associated with a set of image points, which makes all the contributions of camera parameters explicit. We then give briefly some examples of analytical stability results in very simple cases. Since any further theoretical analysis appears extremely difficult, we then present a purely experimental study of the problem. After having given a few conclusions taken from a simulation step, we select and comment some results of the experimental approach based on a robot/camera testbed. We observe an apparently good robustness of the method. General comments on the approach and guidelines for future work are given in the conclusion.