A Continuous Moving Viewpoint Formulation With Joint Optimisation for Efficient Coverage Path Planning in High-Rise Buildings

Abstract

High-rise building inspection through drones requires designing the shortest path around the building while ensuring complete coverage, known as coverage path planning. Conventional approaches select viewpoints from precomputed sets and route them afterwards, fixing viewpoints before the path is known. We propose a geometry-aware moving-viewpoint formulation where each viewpoint samples positions on its local tangent plane and keeps the best under a joint coverage and path objective. At matched coverage on six buildings, we report shorter tours than a multi-offset greedy baseline in 23 of 24 comparisons, by 5–9% at a 90% coverage target. The gap narrows as coverage rises on five buildings but widens to 10% on the most complex. A path-aware greedy baseline finds shorter tours on most buildings using more viewpoints, but loses most runs once a per-stop cost is counted. Adding particle swarm optimisation degrades the solution, which we attribute to missing per-viewpoint credit assignment.