Adaptive UAS Route Planner Based Upon Evidence Accrual

摘要

With the continual increase in governmental and commercial uses of unmanned aerial systems (UASs), flight planning systems that provide both an initial flight plan and a flight plan that is modified as current information is provided become necessary. With UAS operations, flight paths will need to be able change rapidly as restricted airspace can vary quickly, with flight traffic that can increase at a moment's notice, and due to environmental conditions, that can adversely affect the performance of a UAS, i.e., wind corridors in downtown areas. Planning algorithms are needed that better take into account onboard sensor reports, information from off-board reports such as weather and traffic patterns, and information regarding known fixed locations, such as maps of buildings, restricted space, and GPS-dark zones. A technique that was developed for military replanning provides the core for a computationally lightweight and rapidly adaptive UAS route planning. The proposed UAS planning system is described that can offer adaptations from the current plan in response to observations in flight, based upon game-theoretic assessment. The system overlays a map with a hex grid and ranks each hex using evidence accrual based on its effect on the UAS to achieve its goal. The evidence accrual system generates both a score and uncertainty. The data is incorporated using a fuzzy Kalman filter and then is fused via a linear systems-theory approach. With both a score and an uncertainty, the plan can be weighted as possible solution based on achieving the goal of UAS (e.g, delivering a package and returning to base) with a good probability of being correct. As new information is provided, the planning system can change its route based on its current location. This effort includes a description of the route planning system and subcomponents. A simulation of the system is provided with a variety of conditions that modify the original plan both from the start and while in transit.