UAV Path Planning for Building Structure Inspection

Automated path planning for Unmanned Aerial Vehicles (UAVs) is crucial for efficient and safe building inspection, providing a superior alternative to hazardous manual methods. Although generating a flight path from a 3D model is a standard approach, achieving complete surface coverage while adapting to varying geometric complexities remains challenging. This paper introduces a modular two-stage framework to address these issues. The first stage, Viewpoint Generation, systematically produces high-quality viewpoints that guarantee coverage, incorporating a novel normal vector correction mechanism to handle complex geometries. The second stage, Path Planning, adopts an adaptive dual-strategy: a layered MST-DFS method for rapid path generation in structured environments, and an A*-based local replanner for higher path quality in cluttered spaces. The framework was validated through high-fidelity software-in-the-loop simulations in AirSim across multiple real-world building models. Experimental results show that the MST-DFS method achieves over a fivefold reduction in path generation time compared to A*, while A* yields paths up to 14% shorter. Both methods maintained a 100% collision-free success rate. These findings demonstrate that the proposed dual-strategy framework enables mission-specific trade-offs between planning efficiency and path quality, making it a practical and robust solution for automated UAV inspection in diverse indoor environments.