Generative Model-Driven Sampling Strategy for the High-Efficiency Measurement of Complex Surfaces on Coordinate Measuring Machines

Abstract

Coordinate measuring machines are widely used in the precision measurement of manufacturing workpieces. However, the nature of a point-by-point probing characteristic limits their efficiency in the measurement of complex parts, which normally requires dense sampling points for fully evaluating machining errors with high fidelity. To address this problem, this article proposes a generative model-driven sampling strategy to reduce the number of sampling points while preserving the measurement accuracy. Specifically, the reconstruction of surface errors with sparse sampling is transformed as a point cloud super-resolution task, which constructs a generative model to estimate accurate dense results from sparse sampled data. A multiscale neural network architecture is designed to achieve reconstruction, and the fractional Brownian motion is applied to simulate machining errors and synthesize a large-scale dataset for model training. The generalized neural model can then utilize sparse measurements to reconstruct global machining errors, which dramatically reduces the sampling time and increases measurement efficiency. Both computer simulations and actual measurements are carried out to verify the effectiveness of the proposed method.