A Scalable Hierarchical Path Planning technique for Autonomous Inspections with multicopter drones

Multicopter drones equipped with cameras can perform rapid inspections of large buildings, including those with hard to reach features, like bridge pylons. Drones can be made autonomous by providing them with a method to choose a path that maximizes the collected information during the limited flight time allowed by the battery. It is therefore crucial to optimize the trajectories to minimize inspection time. The problem of finding an approximately optimal path passing through a series of desired inspection points in a three-dimensional environment with obstacles is considered. A hierarchical approach is proposed, where the space containing the inspection points is partitioned into different regions and multiple instances of the TSP (Travelling Salesman Problem) are solved, decreasing the overall complexity. An extended graph is used in the TSP, in order to tackle the problem of collision avoidance while planning the trajectory between point pairs. This approach leads to an efficient and scalable method capable of avoiding obstacles, and significantly reduces the time needed to find an optimal path with respect to non-hierarchical methods. Simulation results highlight these features.