The Advantage of Mounting a Camera onto a Robot Arm

Abstract

In this paper we describe a method for recovering Euclidean structure of an unknown scene with an uncalibrated camera mounted onto a robot arm. More precisely the method achieves the following tasks: the projective structure of the scene obtained with a projective reconstruction algorithm is converted into an Euclidean structure and the intrinsic camera parameters are estimated as well as the hand/eye calibration parameters. We cast the problem into a simple set of matrix equations and we determine under which motions the solution is unique. Then we provide a simple algorithm for implementing the method in practice. Finally the experimental results obtained with our method are compared with classical camera and hand/eye calibration techniques. 1 Introduction One of the most challenging tasks in Computer Vision is the task of recovering the 3-D Euclidean structure of a scene with a moving camera. Traditionally there are two possible approaches to this problem. The first approach uses a...