Abstract:In the online assembly process of auto parts,in order to reduce the assembly error caused by the deviation between the theoretical digital model and the actual position,an improved B spline hierarchical interpolation algorithm is designed.First,the initial data is layered according to the curvature distribution of the reconstructed surface of the point cloud,and then the deviation distribution of the actual part point cloud and the digital analog point cloud is obtained through the cubic B spline interpolation algorithm.Finally,the path compensation is derived from the absolute distance deviation value normalization coefficient,so as to realize the optimization of scanning path for auto parts detection.The parts containing right angles,edges and holes are analyzed through the simulation,and the results show that the denoising effect of the point cloud is good,and the surface shape of the reconstructed surface conforms to the characteristics of the solid part.In the experiment,the reconstruction effect of direct reconstruction and the optimization of the cubic B spline hierarchical interpolation are compared .Through the statistics of 300 test points,the average errors of the three coordinate axis directions before and after optimization are 3.345mm and 0.599mm,respectively.It is demonstrated that the position prediction accuracy has been improved after optimization,laying a foundation for improving the accuracy of the scanning path.