Learning Transferable Cooperative Behavior in Multi-Agent Teams

TL;DR

This paper introduces a graph-based multi-agent reinforcement learning framework that enables agents to learn cooperative behaviors, generalize to different team sizes, and operate effectively in decentralized, real-world scenarios.

Contribution

The paper presents a novel shared agent-entity graph approach for multi-agent cooperation that is invariant to team size and permutation, with state-of-the-art results and strong zero-shot transfer capabilities.

Findings

Achieved state-of-the-art results on coverage, formation, and line control tasks.

Demonstrated rapid transfer to different team sizes with zero-shot generalization.

Operates effectively in fully decentralized settings.

Abstract

While multi-agent interactions can be naturally modeled as a graph, the environment has traditionally been considered as a black box. We propose to create a shared agent-entity graph, where agents and environmental entities form vertices, and edges exist between the vertices which can communicate with each other. Agents learn to cooperate by exchanging messages along the edges of this graph. Our proposed multi-agent reinforcement learning framework is invariant to the number of agents or entities present in the system as well as permutation invariance, both of which are desirable properties for any multi-agent system representation. We present state-of-the-art results on coverage, formation and line control tasks for multi-agent teams in a fully decentralized framework and further show that the learned policies quickly transfer to scenarios with different team sizes along with strong zero-shot generalization performance. This is an important step towards developing multi-agent teams which can be realistically deployed in the real world without assuming complete prior knowledge or instantaneous communication at unbounded distances.