Collaborative Auto-Curricula Multi-Agent Reinforcement Learning with Graph Neural Network Communication Layer for Open-ended Wildfire-Management Resource Distribution

Most real-world domains can be formulated as multi-agent (MA) systems. Multiple 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 positive 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 self-performance or send help requests for predicted increased task difficulty. We conduct our study in the context of resource distribution for wildfire management. Communicating environmental features and partially observable fire occurrence observations help the agent collective pre-emptively distribute resources. Furthermore, we introduce a procedural training environment accommodating auto-curricula and open-endedness towards better generalizability. While our MA communication proposal includes a large action and observation space, we outperform a Greedy Heuristic Baseline, a Single-Agent (SA) and MA setup.