Learning Communication for Decentralized Coordination in Multi-Robot Systems

Effective communication is key to successful, decentralized, multi-robot coordination. Yet, it is far from obvious what information is crucial to the task at hand, and how and when it must be shared among agents. This talk covers using Graph Neural Networks to solve multi-robot coordination problems — from near-optimal decentralized multi-agent path finding, to learning inter-agent communication policies via GNN-based reinforcement learning, to the challenge of sim-to-real transfer.