Pay-Per-Flight Dynamic Pricing of UAV Operations

Abstract

Insuring unmanned aerial vehicles (UAVs) is a relatively new concept, where not much data is available yet. We propose the combination of available data from different sources, other than past accident rates, to stochastically model the operational environment by using Gaussian process-based function approximations. A data-driven risk measure is then derived through such stochastic formulation accounting for both aleatoric uncertainties of the considered environmental factors as well as epistemic uncertainties originating from the geographical sparsity of data collection sources. The risk measure is obtained in a path-integral form which represents the operational risk associated with a defined operation in partially unknown environments. A novel pay-per-flight dynamic pricing scheme is derived from such risk measure.