A fault-tolerant automatic mission planner for a fleet of aerial vehicles

Abstract

Unmanned Aerial Vehicles are versatile tools for inspection tasks, for example of the built environment. Due to their limited flight time, it is sensible to employ multiple units to decrease the total mission time and avoid additional trips to change batteries. The use of more units makes the mission planning more complex; moreover, a fault tolerant plan that is resilient at least to a single failure is desirable. To solve these problems, a new algorithm is proposed, which, via hierarchical decomposition and numerical optimization, effectively deals with: (1) the efficient generation of a suitable path for each drone, and (2) guaranteeing mission robustness against a single fault. The automated generation and clustering of points of interest to be visited is addressed, too, as part of the whole procedure. Using accurate models of two real buildings, it is shown that the approach delivers close-to-optimal solutions with small computational time, thus being compatible with real-world operation.