A Cognitive Framework for Dynamic Level-of-Autonomy Selection in Autonomous UAV Missions

The increasing deployment of Unmanned Aerial Vehicles (UAVs) in surveillance, inspection, monitoring, and mission-critical applications calls for autonomous systems capable of adapting their Level of Autonomy (LoA) to dynamically changing operational conditions. Conventional UAV autonomy schemes often rely on predefined configurations or isolated threshold-based rules, which may provide limited adaptability when environmental, communication, system, mission, and human-operator conditions change concurrently. The present study proposes a framework based on cognitive decision-making, enabling UAVs to dynamically determine the most appropriate LoA during mission execution. The framework combines Bayesian data-driven reasoning with event-driven and anticipatory reasoning to jointly assess environmental complexity, weather conditions, obstacle risk, mission criticality, UAV health, remaining energy, navigation confidence, communication quality, computational resources, operator workload, availability, experience, and preferred autonomy. The framework is evaluated through simulations covering nominal operation, communication degradation, increased operator workload, unexpected obstacle detection, weather deterioration, UAV-health degradation, and combined multi-disturbance scenarios. The proposed approach provides a unified cognitive mechanism for context-aware and event-driven UAV autonomy adaptation, offering a flexible foundation for resilient and human-centered autonomous UAV operations.