Information-Theoretic Optimization of Periodic Orbits for Persistent Cooperative Geolocation
William W. Whitacre, George Tillinghast, Mark Campbell · AIAA Guidance, Navigation, and Control Conference and Exhibit · 2008
The optimization of periodic orbits for tracking both stationary and moving targets with two uninhabited aerial vehicles with camera sensors is considered. A natural informationtheoretic metric is defined which takes into account the fact that the orbits are periodic. In keeping with the current literature, the search space of orbits is restricted to circular orbits of constant radius and altitude. However, additional freedom is created by allowing the orbits to be offset, as opposed to centered about the target. Three key results are obtained from the optimization. First, the widely held belief that aircraft in target centric orbits should be separated in their orbits by 90 ◦ is shown to be correct only when the target altitude is known. Second, allowing the orbits to be offset from the target can provide a significant increase in tracking performance. Third, the level of process noise in the target model affects the optimal choice of orbit parameters. These optimization results are used to evaluate the use of approximate metrics that are based on instantaneous information measures. These approximate metrics, though common in the literature, are shown both analytically and numerically to be limited in their range of applicability. Also, simulations of the ScanEagle UAV for stationary and moving targets were performed for various aircraft phase angle commands and orbit offsets. These simulations are used to develop a number of practical guidelines for persistent cooperative geolocation.