Coverage path planning for structural inspections with UAV

Abstract

This study presents a comprehensive framework for complete inspection of known structures using an Unmanned Aerial Vehicle subject to position uncertainty. The proposed strategy is to take a two-stage approach consisting of viewpoint planning followed by routing. The proposed method enhances an existing Viewpoints Generation algorithm, ensuring 100% coverage of the structure while significantly reducing both computing time and the number of viewpoints by up to 99% and 28%, respectively. The path planning problem is formulated to optimise either energy consumption or mission duration. Various Travelling Salesman Problem (TSP) solvers are compared, with the Lin-Kernighan Heuristic providing near-optimal results in a short time frame when applied to a full-scale wind turbine model. Validation is conducted in the Gazebo simulation environment, offering realistic conditions, while PX4 Software-In-The-Loop (SITL) and ROS2 ensure seamless deployment on real hardware. The code is available in the Zenodo archive [11].