The Contour Error Evaluation Based on Spherically-Approximated Patches for Free Form Surface

Wei Fan,Kang Min,Houfu Dai,Zhiwei Su
DOI: https://doi.org/10.1007/s00170-024-14161-y
2024-01-01
Abstract:The contour error evaluation in computer-aided manufacturing (CAM) systems is critical for high-precision measurement and manufacture of free-form surfaces. The existing methods of evaluating the contour error for free-form surfaces have been developed based on various models. Nevertheless, there still exist some of the shortcomings when applied practically, including the inaccurate distance model between the measured point cloud (MPC) and ideal surface, the numerical search required for the correspondence, possible local optimum of the solution, the inapplicability to all the surfaces, the huge point cloud required, and heavy computational load in the registration iteration, and so forth. To overcome those shortcomings, this work proposes the registration method based on a spherically approximated patch (SAP) for the contour error evaluation of free-form surfaces. The significance of registration between the MPC and the SAP set is theoretically analyzed. Then the spherically approximable condition is provided, and the SAP set for the ideal surface is efficiently generated based on the centroid-midpoint search of parametric triangle subdivision (CMSPTS), and the meshes disobeying the spherically approximable condition are refined to planar patches under prescribed tolerance. The tactics of efficiently computing the projection of the MPC on the SAP set based on a local spreading search function (LSSF) are provided. Finally, the scheme of evaluating contour error based on the Point-To-SAP (PTSAP) registration is developed. The comparative simulation and experiment demonstrate that the proposed method raises the registration level between the MPC and the ideal surface, produces the evaluation result closer to the theoretical one with acceptable efficiency, and reduces the size of the ideal point cloud compared with the existing methods.
What problem does this paper attempt to address?