Into the Unknown: UAV Navigation with F-Star (F*)

With increasing demand for flexible aerial systems, Unmanned Aerial Vehicles (UAVs) have become central to research in areas like emergency response, mobile networking, and surveillance. Multirotor UAVs enable seamless movement in all directions, but introduce challenges in localization, mapping, and real-time path planning. Modern missions require more than basic collision avoidance or shortest-path computation; they demand energy-efficient, safe, and adaptive navigation. In networking applications, additional constraints include restricted airspaces, frequency zoning, radio silence mandates, and uplink visibility. Although UAV path planning has been studied extensively, many existing methods are computationally intensive. To address this, we propose F-Star (F*), a framework that integrates adaptive spatial partitioning, weighted graph search, and trajectory smoothing for efficient and safe UAV navigation, while providing extensibility for multi-objective planning. A preliminary study in realistic scenarios demonstrates our proposed solutions effectiveness in dynamic obstacle avoidance with low computational cost and extensibility for multi-objective planning domains.