Flight path planning of UAV-driven refinement inspection for construction sites based on 3D reconstruction

Construction site inspection usually requires high labor demands and remains inefficiency. Current UAV-based inspection mainly relies on manual operation with challenging low-altitude path in complex site environment. This paper proposes an automatic inspection flight path planning method based on BIM, UAVs, and 3D reconstruction for autonomous site inspection. First, the UAV performs rough flights to reconstruct a 3D model of the construction site, which, combined with BIM, enables environmental perception. Then, waypoint spaces are generated by mesh expansion, and an improved artificial potential field method combined with a greedy algorithm determines inspection waypoints and headings. Next, A* and simulated annealing algorithms are used to compute a globally optimal path. Finally, local coordinates are converted to real-world coordinates to generate UAV flight files and complete the autonomous inspection task. Experiments demonstrate enhanced safety and efficiency, with a 20.3 % improvement in inspection speed and reduced battery usage from 1.51 to 1.32 units.