Hierarchical Lead Critic based Multi-Agent Reinforcement Learning
This work addresses coordination problems in multi-agent systems, offering a novel approach that enhances scalability and robustness, though it is incremental in building on existing hierarchical methods.
The paper tackles the challenge of cooperative multi-agent reinforcement learning by introducing a hierarchical lead critic architecture that combines local and global perspectives, resulting in improved performance and sample efficiency across various benchmarks.
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.