Decentralized Distribution of UAV Fleets Based on Fuzzy Clustering for Demand-driven Aerial Services

Unmanned aerial vehicles provide a flexible solution for emerging civilian applications such as last-mile delivery, remote environmental monitoring or hazard detection. However, since the demand for this type of applications is sparsely distributed and comprises spatiotemporal uncertainty, prior to the deployment of aerial means, the dimension and configuration of multi-UAV systems has to be addressed. This paper proposes a clustering-based dimensioning strategy for networks of aerial vehicles, that incorporates heterogeneous vehicle characteristics and energy-related autonomy constraints to determine the type and number of vehicles required to meet a given demand. The approach follows a fuzzy data-driven method which enables generating coverage regions with elastic boundaries that allows for cooperation between neighboring regions to manage high and low demand scenarios, providing improved resilience to demand fluctuations and guaranteeing a well-conditioned formulation for the optimal resource allocation problem.