Position-based visual servoing through Cartesian path-planning for a grasping task

In this paper three visual servo schemes using the image-based and position-based techniques in a decoupled fashion are presented. These schemes are applied to grasping objects with orientation and position errors in an industrial grasping task. The servoing schemes are evaluated according to their rate of convergence to the desired grasping pose. The first two schemes rely on a constant image features extraction while the third one bypasses this reliance and uses a pose estimation of the camera at the initial and desired positions with a control law designed in 3D Cartesian space by Path Planning. It can be shown that the latter method ensures the shortest convergence time provided that a perfect pose estimation is performed.