An Improved Robotic Grinding Trace Planning Method Based on Projection from 2D Plane onto 3D Surface

With the advent of the intelligent manufacturing,industrial robots have been widely used, e.g., the application of grinding robots significantly improve grinding efficiency and quality. This paper proposes an improved projection method for creating grinding paths on the complex 3D surface of workpiece. It “wraps” a 2D plane with 2D traces on it onto the 3D target surface, then the 2D traces become the 3D curves on the 3D surface, where the 3D curves are the grinding trajectory and maintain the major features of the raw 2D traces. A 3D depth camera is utilized to obtain point cloud of the workpiece surface, which is then filtered and segmented. After estimating the tangent plane of the 3D surface of the workpiece, the normal vector of that position can be determined. Thus the pose of the grinding head can be aligned with the normal vector, which ensures the correct grinding of the surface. When all normal vectors along the 3D curve are estimated, the complete grinding trajectory can be obtained. In order to achieve compliant grinding with high quality, an admittance control algorithm is utilized. Experimental results demonstrate that the proposed projection method can generate the desired grinding trajectory, and the admittance controller help to maintain the contact force and guarantee the grinding accuracy.