Real-time Estimation of UAS Performance Using Efficient Sampling of Functional Models

Numerica proposes to developed advanced algorithms for constructing a UAS vehicle model from ATC surveillance data in real-time. Using functional descriptions of aircraft performance and computationally efficient sampling techniques, UAS model parameters are estimated, and aircraft maneuvers that best improve these estimates are determined. Numerica's approach has two important advantages. First it is agnostic to the specific structure of the aircraft performance model and can be used with a range of parameterized modeling techniques ranging from simplified "table look-up" models to physics-based kinetic models. This permits faster integration with current ATM systems since it leverages (rather than replaces) existing trajectory prediction techniques and databases. Second, the functional descriptions do not require that the models for propulsion or aerodynamic forces to adhere to a specific analytical form. This allows tremendous modeling flexibility and permits the inclusion of complicated atmospheric factors that may be relevant for trajectory prediction but are difficult to capture with simple closed-form expressions. Since the approach has solid theoretical foundations, the algorithms can be also used in an offline context to help establish bounds on the "best-possible" model estimation performance given the accuracy and character of available surveillance data. This capability could help in determining requirements on ATC sensors to enable reliable trajectory predictions for UAS that lack detailed performance models.