Deep Multi-Agent Reinforcement Learning with Discrete-Continuous Hybrid Action Spaces

TL;DR

This paper introduces two novel deep multi-agent reinforcement learning algorithms designed for hybrid action spaces, demonstrating superior performance on complex simulated tasks compared to existing methods.

Contribution

The paper proposes Deep MAPQN and Deep MAHHQN algorithms to address multi-agent DRL with hybrid action spaces, a previously unexplored problem.

Findings

Both algorithms outperform existing independent deep parameterized Q-learning.

Effective in complex tasks like RoboCup Soccer and Ghost Story.

Show significant improvements in training efficiency and policy effectiveness.

Abstract

Deep Reinforcement Learning (DRL) has been applied to address a variety of cooperative multi-agent problems with either discrete action spaces or continuous action spaces. However, to the best of our knowledge, no previous work has ever succeeded in applying DRL to multi-agent problems with discrete-continuous hybrid (or parameterized) action spaces which is very common in practice. Our work fills this gap by proposing two novel algorithms: Deep Multi-Agent Parameterized Q-Networks (Deep MAPQN) and Deep Multi-Agent Hierarchical Hybrid Q-Networks (Deep MAHHQN). We follow the centralized training but decentralized execution paradigm: different levels of communication between different agents are used to facilitate the training process, while each agent executes its policy independently based on local observations during execution. Our empirical results on several challenging tasks (simulated RoboCup Soccer and game Ghost Story) show that both Deep MAPQN and Deep MAHHQN are effective and significantly outperform existing independent deep parameterized Q-learning method.