引用本文:吕梦欣,石宗英,钟宜生.多智能体强化学习在微分博弈中的应用[J].控制理论与应用,2025,42(11):2165~2178.[点击复制]
LV Meng-xin,SHI Zong-ying,ZHONG Yi-sheng.Applications of multi-agent reinforcement learning in differential games[J].Control Theory & Applications,2025,42(11):2165~2178.[点击复制]
多智能体强化学习在微分博弈中的应用
Applications of multi-agent reinforcement learning in differential games
摘要点击 2296  全文点击 198  投稿时间：2025-03-29  修订日期：2025-10-16
查看全文  查看/发表评论  下载PDF阅读器   HTML
DOI编号  10.7641/CTA.2025.50128
  2025,42(11):2165-2178
中文关键词  强化学习  微分博弈  多智能体系统  最优控制
英文关键词  reinforcement learning  differential games  mult-agent systems  optimal control systems
基金项目  国家自然科学基金项目(U24B20173,62573016),国家重点研发计划项目(2020YFF0400100)资助.
作者单位E-mail
吕梦欣 清华大学自动化系 lvmx23@mails.tsinghua.edu.cn 
石宗英 清华大学自动化系 szy@mail.tsinghua.edu.cn 
钟宜生* 清华大学自动化系  
中文摘要
      微分博弈是一种研究对抗环境下多智能体系统决策问题的方法,在经济、工程等领域都有着广泛的应用前 景. 然而当问题建模接近真实场景时,微分博弈的理论求解面临诸多困难.近年来,随着人工智能技术的快速发展, 多智能体深度强化学习方法的研究取得了瞩目突破,为解决微分博弈所面临的挑战提供了新的思路.本文首先对 这二者各自的原理、发展现状进行了梳理;之后详细介绍应用于微分博弈的强化学习算法,并根据微分博弈问题类 别和网络模型进行细分,总结目前研究对传统方法所面临困难的解决情况;最后,对微分博弈与多智能体强化学习 的发展现状进行分析,给出未来可能的研究方向.
英文摘要
      Differential games offer a powerful framework for modeling and analyzing the decision-making problems in multi-agent systems under competitive environments, with extensive application prospects in fields like economics and industry. However, as the problem modeling approaches real-world scenarios, obtaining theoretical solutions becomes increasingly challenging. In recent years, with the rapid advancement of artificial intelligence, multi-agent deep rein forcement learning has achieved significant breakthroughs, providing promising alternatives for addressing the challenges encountered in differential games. This paper firstly provides a comprehensive review of the fundamental principles and lat est development of both fields. The paper further details the reinforcement learning methods applied to differential games, categorizing them based on the problem formulations and deep-learning network architectures, and illustrates how recent researches overcome the challenges faced by traditional methods. Finally, the paper summarizes the current progress in this interdisciplinary research area and suggests potential future directions.