Force Control and Visual Servoing Using Planar Surface Identification

When designing flexible multi-sensor based robot systems, one important problem is how to combine the measurements fromdisparate sensors such as cameras and force sensors. In this paper,we present a method for combining direct force control and visualservoing in the presence of unknown planar surfaces. The controlalgorithm involves a force feedback control loop and a vision basedreference trajectory as a feed-forward signal. The vision system isbased on a constrained image-based visual servoing algorithm designedfor surface following, wherethe location and orientation of the planar constraint surface is estimated online using position-, force- and visual data.We show how data from a simple and efficient camera calibration method can be used in combination withforce and position data to improve the estimation and referencetrajectories.The method is validated through experiments involving force controlleddrawing on an unknown surface. The robot will grasp a pen and use itto draw lines between a number of markers drawn ona white-board, while the contact force is kept constant. Despite its simplicity, the performance of the method is satisfactory.