Digital twin-enabled error and uncertainty mapping for 3D scanning

3D scanners have great potential to boost the efficiency of geometric inspection if their traceability is established by identifying measurement bias and uncertainty. This article introduces a digital twin for a structured-light 3D scanner that provides these two essential measures in a real-time manner. The error quantification starts from iterated scans taken from a CMM-calibrated ball-plate artifact at eight different configurations. The differences between the erroneous scan readings of the artifact's distance features and the corresponding calibrated values in the CMM set give a spatial distribution of the scanner's error. Then, three polynomial functions map the Cartesian coordinates of the scan set onto their counterparts from the CMM set. In this error model, a least-square approach determines the polynomial coefficients. Evaluating the error model for a given point cloud returns a vector field of volumetric errors, which is then followed by error corrections and then GD&T calculations. The uncertainty propagation is pursued through the three steps of model development, error map evaluation, and GD&T analysis. The Monte-Carlo estimations benchmark and validate the GUM framework for the uncertainty associated with the outputs of each step. Experimental verifications of the proposed methodology suggest that the error correction improves the mean and the standard deviation of measurement bias by 50% and 27%, respectively. The uncertainty map provides the coverage probabilities of 62%, 88%, and 96% for the coverage factors of 1, 2, and 3, while the expected theoretical levels of confidence are 68%, 95%, and 100%.