Point cloud 3D parent surface reconstruction and weld seam feature extraction for robotic grinding path planning

High-performance components with complex geometries make it difficult to determine the position and orientation of grinding tool. In this work, a fast and accurate robotic grinding path planning method is proposed for automatic removal of irregular weldments on a free form surface. The surface of workpiece is digitalized by 3D profile scanner and represented by point cloud data. Statistic filter, weighted least square regression and differences of normal vectors are used for point cloud pre-processing and segmentation. All segments are then modelled by B-spline surfaces to obtain the parent surface. A new superposition method is proposed to create a computer-aided design (CAD) model of the actual workpiece by adding the weld seam to the parent surface. The CAD model is then imported into an off-line simulation system to generate and execute grinding path. With the superposition method, the heights and widths of weld seam are extracted by analysing the difference between point cloud data and the reconstructed parent surface in order to determine the feed rate and size of grinding tool. Experimental results show that the proposed superposition method has the maximum absolute percentage error 5.3% and 41% saving in computation time in comparison with the conventional reverse engineering method.