Capturing Reality Changes from Point Clouds for Updating Road Geometric Digital Twins

Abstract: Proactive maintenance of roads enables increased asset lifespan with improved safety and minimised downtime. However, the absence of up-to-date structured information leads to expensive reactive maintenance. Geometric digital twins (GDT) provide digital replicas of object geometry, but no automatic tools exist to maintain them. This paper develops a method for detecting and applying geometric changes to road GDTs. Our solution performs a distance-based comparison of point clouds, estimates the changes and then applies them to GDT. It achieves a 77.87 F1-score in detecting changes and significantly saves time by automating the process, making such digital twins viable and practically applicable.