This paper proposes a concept of navigation and guidance strategy planner for urban operation of a VTOL-type UAV. One of major challenges of UAV autonomous navigation in an urban environment is to deal with the risk of GPS signal occlusion. In order to address this issue, various approaches have been proposed for GPS-independent UAV navigation and guidance such as visual odometry and visual servoing control. In this context, this work supposes that different navigation and guidance modes using different set of sensors are available onboard an UAV. An idea of the proposed planner is to anticipate the navigation and guidance performance degradation (or amelioration) due to unavailability (or availability) of certain mode in the path planning task. The planning problem is formulated as a 5D (3D position + selection of navigation and guidance modes) graph search problem, where the localization and path execution uncertainties are propagated according to a model of corresponding modes for each node transition. Node transition is denied if the path execution uncertainty ellipsoid intersects with any obstacle. A minimizing cost function is defined by a volume of the path execution uncertainty corridor, as it implies minimizing path distance and execution accuracy at the same time. A deterministic graph search algorithm is applied to find a flight path with specified navigation and guidance mode transitions which minimizes the defined cost function. Simulations are performed by using path planning configurations given in an existing UAV obstacle field navigation benchmark, and the results are presented to prove the proposed navigation and guidance strategy planning concept.
展开▼
