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

arXiv cs.AI · 6 hours ago

Researchers developed a three-level learning architecture for autonomous UAV swarms in search and rescue that combines Hebbian learning, multi-agent reinforcement learning with graph neural networks, and model-agnostic meta-learning. The framework formalizes 22 architectural contracts across six components that provide six classes of formal guarantees including safety, optimality, and liveness. The hybrid neuro-symbolic system enables swarms to monitor their own cognitive state and switch between strategies, addressing limitations of existing hierarchical reinforcement learning approaches.