Collaborative Auto-Curricula Multi-Agent Reinforcement Learning with Graph Neural Network Communication Layer for Open-ended Wildfire-Management Resource Distribution
This work addresses a domain-specific problem in wildfire management by enabling agents to share information and coordinate actions, though it is incremental as it builds on existing MARL and GNN methods.
The paper tackles the challenge of teaching multi-agent systems to collaborate effectively in resource distribution for wildfire management, achieving improved performance over greedy heuristic and single-agent baselines through a MARL system with GNN communication and auto-curricula training.
Most real-world domains can be formulated as multi-agent (MA) systems. Intentionality sharing agents can solve more complex tasks by collaborating, possibly in less time. True cooperative actions are beneficial for egoistic and collective reasons. However, teaching individual agents to sacrifice egoistic benefits for a better collective performance seems challenging. We build on a recently proposed Multi-Agent Reinforcement Learning (MARL) mechanism with a Graph Neural Network (GNN) communication layer. Rarely chosen communication actions were marginally beneficial. Here we propose a MARL system in which agents can help collaborators perform better while risking low individual performance. We conduct our study in the context of resource distribution for wildfire management. Communicating environmental features and partially observable fire occurrence help the agent collective to pre-emptively distribute resources. Furthermore, we introduce a procedural training environment accommodating auto-curricula and open-endedness towards better generalizability. Our MA communication proposal outperforms a Greedy Heuristic Baseline and a Single-Agent (SA) setup. We further demonstrate how auto-curricula and openendedness improves generalizability of our MA proposal.