Multi-Agent Reinforcement Learning with a Hierarchy of Reward Machines

TL;DR

This paper introduces MAHRM, a hierarchical reward machine approach for cooperative multi-agent reinforcement learning, enabling efficient learning in complex, interdependent scenarios by leveraging high-level event hierarchies.

Contribution

It presents a novel hierarchical reward machine framework for MARL that handles complex, interdependent events, improving learning efficiency over existing methods.

Findings

MAHRM outperforms other MARL methods in three domains.

Hierarchical decomposition reduces computational complexity.

Handles concurrent high-level events among agents.

Abstract

In this paper, we study the cooperative Multi-Agent Reinforcement Learning (MARL) problems using Reward Machines (RMs) to specify the reward functions such that the prior knowledge of high-level events in a task can be leveraged to facilitate the learning efficiency. Unlike the existing work that RMs have been incorporated into MARL for task decomposition and policy learning in relatively simple domains or with an assumption of independencies among the agents, we present Multi-Agent Reinforcement Learning with a Hierarchy of RMs (MAHRM) that is capable of dealing with more complex scenarios when the events among agents can occur concurrently and the agents are highly interdependent. MAHRM exploits the relationship of high-level events to decompose a task into a hierarchy of simpler subtasks that are assigned to a small group of agents, so as to reduce the overall computational complexity. Experimental results in three cooperative MARL domains show that MAHRM outperforms other MARL methods using the same prior knowledge of high-level events.