Hybrid Control of a DD Robot Using an Online Visual and Force Sensor Fusion Method

Abstract

In this paper a force and vision sensor fusion method is proposed in order to extend the conventional force and position hybrid control. The purpose of this system is to apply a controlled normal force on a surface with unknown shape. We achieve our objectives by using a partially reconstructed trajectory, visualized by one CCD camera installed on the end-effector, and complementing it with the tactile information provided by a force sensor. Based on the force sensor information, the normal and tangential directions of the constraint surface are identified and the desired point, obtained from the CCD camera, is projected back to this surface. The only assumption done here is that the surface is smooth enough to allow us to do a good estimation of the constraint surface. This method was developed in order to simplify the use of the 6-DOF direct-drive (DD) robot proposed by Utsumi and Todo, which was initially designed to do polishing tasks. The referred DD robot was designed with a particular structure in order to lower its whole weight : a 3-DOF parallel link arm robot and a 3-DOF orientation table. By using this system, the user can easily draw an arbitrary path on the surface of the object before he fixes it on the orientation table, without thinking about complicated curved path and surface equations or coordinate system transformations.