Hierarchical Lead Critic based Multi-Agent Reinforcement Learning

TL;DR

This paper introduces a Hierarchical Lead Critic architecture for multi-agent reinforcement learning, enabling agents to learn from multiple hierarchical perspectives, improving performance, sample efficiency, and scalability in complex cooperative tasks.

Contribution

The paper proposes a novel hierarchical MARL architecture with a sequential training scheme, combining local and global perspectives for better coordination.

Findings

HLC outperforms single hierarchy baselines in various benchmarks.

HLC scales robustly with more agents and task difficulty.

HLC achieves high sample efficiency and robust policies.

Abstract

Cooperative Multi-Agent Reinforcement Learning (MARL) solves complex tasks that require coordination from multiple agents, but is often limited to either local (independent learning) or global (centralized learning) perspectives. In this paper, we introduce a novel sequential training scheme and MARL architecture, which learns from multiple perspectives on different hierarchy levels. We propose the Hierarchical Lead Critic (HLC) - inspired by natural emerging distributions in team structures, where following high-level objectives combines with low-level execution. HLC demonstrates that introducing multiple hierarchies, leveraging local and global perspectives, can lead to improved performance with high sample efficiency and robust policies. Experimental results conducted on cooperative, non-communicative, and partially observable MARL benchmarks demonstrate that HLC outperforms single hierarchy baselines and scales robustly with increasing amounts of agents and difficulty.