Hierarchical Cooperative Multi-Agent Reinforcement Learning with Skill Discovery
This work addresses the challenge of creating intelligent agents for fully cooperative multi-agent environments, such as team sports games, with potential for human-AI cooperation, though it is incremental as it builds on existing hierarchical and skill discovery methods.
The paper tackles the problem of achieving high-level coordination in cooperative multi-agent settings by proposing a two-level hierarchical MARL algorithm with unsupervised skill discovery, resulting in the emergence of useful and distinct skills and cooperative team play in experiments on a stochastic high-dimensional team game.
Human players in professional team sports achieve high level coordination by dynamically choosing complementary skills and executing primitive actions to perform these skills. As a step toward creating intelligent agents with this capability for fully cooperative multi-agent settings, we propose a two-level hierarchical multi-agent reinforcement learning (MARL) algorithm with unsupervised skill discovery. Agents learn useful and distinct skills at the low level via independent Q-learning, while they learn to select complementary latent skill variables at the high level via centralized multi-agent training with an extrinsic team reward. The set of low-level skills emerges from an intrinsic reward that solely promotes the decodability of latent skill variables from the trajectory of a low-level skill, without the need for hand-crafted rewards for each skill. For scalable decentralized execution, each agent independently chooses latent skill variables and primitive actions based on local observations. Our overall method enables the use of general cooperative MARL algorithms for training high level policies and single-agent RL for training low level skills. Experiments on a stochastic high dimensional team game show the emergence of useful skills and cooperative team play. The interpretability of the learned skills show the promise of the proposed method for achieving human-AI cooperation in team sports games.