Geometric digital twins of long-living assets: uncertainty-aware 3D images from measurement and CAD data

Abstract

Listen

Translate article

Geometric digital twins (gDT) of long-living assets are formed from a multitude of different data sources like CAD or measurement data imaging as-designed to sensory-based acquired as-is states. Change detection techniques allow merging 3D data pairwise, deriving a geometric digital twin of the physical asset. However, positional uncertainties must be considered to account for the data’s reliability when facing different kinds of data sources. The contribution of this work provides the ability to specify (an-)isotropic positional uncertainty quantities for meshes and even sparse point clouds in advance, enabling uncertainty-aware change detection. In the case of sensory-based as-is data, uncertainty is defined by the measurement process, while in the case of as-designed CAD data, the uncertainty relates, e.g., to manufacturing tolerances. Instead of testing on equal means during point-wise comparison as in M3C2, we utilize the Bhattacharyya distance measure to quantify the (dis)similarity between correspondences during change detection.