Intelligent Three Level Learning Architecture for Autonomous UAV Swarms in Search and Rescue

This paper introduces a new three-level hierarchical learning architecture designed for autonomous UAV swarms in search and rescue operations. It integrates distinct learning mechanisms—Hebbian neuroplasticity, multi-agent reinforcement learning with graph neural networks, and model-agnostic meta-learning with BDI reasoning—to enable individual adaptation, tactical coordination, and strategic decision-making. The architecture provides formal guarantees for safety, optimality, and resilience, addressing limitations in existing hierarchical reinforcement learning approaches.