Automatic Programming for Robotic Grinding Using Real Time 3D Measurement

Abstract

In order to apply an industrial manipulator to grind or polish an unknown work piece, many robot tool points must be taught manually. This programming procedure could be complex if the work piece has arbitrary surfaces meanwhile its digital 3D model is unknown. In this paper, an automatic path planning for unknown work piece grinding is proposed using the 3D point cloud of the work piece. The surface defects are firstly recognized, then the relevant 3D point clouds are reconstructed using a real-time structure light based 3D scanner. The grinding path can be generated after the surface fitting using the point cloud. The manipulator force control is applied to keep a uniform grinding force and remedy the error of the surface fitting. We set up a test platform of the automatic grinding system, including a 3D scanner and an industrial manipulator. The experimental result proves that the proposed technique can realize automatic robotic programming of unknown work piece and has the potential for practical applications.