Mixed Reality-based MEP construction progress monitoring: Evaluation of methods for mesh-to-mesh comparison

Visually monitoring progress and geometric quality on site using Mixed Reality (MR) and overlaid Building Information Model (BIM model) is challenging, particularly in complex contexts like complex mechanical, electrical, and plumbing (MEP) systems. This paper proposes and evaluates four individual methods and three combined ones for automated object recognition and deviation evaluation, based on the matching and comparison of the 3D mesh captured on site by MR systems with the mesh geometry of the elements in the (as-designed) BIM model. The four individual methods include: (1) Bounding Box Occupation, (2) Point-to-Surface Distance, (3) Voxel Occupation, (4) Feature Matching. Three combined methods are Method 1 ∪ 4 , Method 2 ∪ 4 and Method 3 ∪ 4 (i.e. combining methods 1 and 4, 2 and 4, and 3 and 4, respectively). The methods are evaluated using both synthetic and real data of MEP construction works, with the Method 1 ∪ 4 yielding the best performance. • Automated detection of MEP elements for construction monitoring utilising MR and BIM. • Four individual and three combined algorithms assessed for efficient mesh-to-mesh comparison. • “Box Occupation + Feature Matching” method yields superior detection + computational results.