Keeping features in the camera's field of view: a Visual servoing strategy

In this paper we propose a new visual servoing technique using point correspondences. This approach ensures, for known intrinsic parameters and optical axis direction, global convergence and guarantees to keep all points in the field of view of the camera during the robot control, without requiring any knowledge on the 3D model of the scene. Moreover, the camera trajectory length is minimized in the rotational space and, for some cases including all the ones in which the features do not reach the image boundary, also minimized in the translational space. Robustness against uncertainties is also guaranteed.