Visual servoing based on an analytical homography decomposition

This paper presents a new vision-based control method for positioning a camera with respect to an unknown planar object. Standard methods use non-linear state observers based on homography decomposition. In the general case, there are two possible solutions to the decomposition problem. Thus, some additional "a priori" information must be used. In this paper, we propose to use an analytical decomposition of the homography matrix in order to define a new control objective that allows to discard the false solution without any "a priori" information. The stability of the control law has been proved.