Specializing Inter-Agent Communication in Heterogeneous Multi-Agent Reinforcement Learning using Agent Class Information
This work aims to improve communication efficiency and performance in multi-agent systems for researchers and practitioners working with heterogeneous agent teams, offering an incremental improvement to existing communication methods.
This paper addresses inter-agent communication in heterogeneous multi-agent reinforcement learning by representing communication capabilities as a directed labeled heterogeneous agent graph. The proposed neural network architecture specializes communication by learning individual transformations for messages exchanged between each pair of agent classes, achieving comparable or superior performance in environments with a larger number of agent classes.
Inspired by recent advances in agent communication with graph neural networks, this work proposes the representation of multi-agent communication capabilities as a directed labeled heterogeneous agent graph, in which node labels denote agent classes and edge labels, the communication type between two classes of agents. We also introduce a neural network architecture that specializes communication in fully cooperative heterogeneous multi-agent tasks by learning individual transformations to the exchanged messages between each pair of agent classes. By also employing encoding and action selection modules with parameter sharing for environments with heterogeneous agents, we demonstrate comparable or superior performance in environments where a larger number of agent classes operates.