Multi-Task Multi-Agent Reinforcement Learning via Skill Graphs

doi:10.1109/LRA.2025.3588784

PRE

Guobin Zhu

DOI：10.1109/LRA.2025.3588784

原文链接

原文求助

摘要

En 中文

多任务多智能体强化学习（MT-MARL）因其能够增强MARL在多个任务上的适应性而 recently gained attention。然而，现有多任务学习方法在处理复杂问题时面临挑战，因为它们无法处理不相关任务，且知识迁移能力有限。本文提出了一种分层方法，能够高效解决这些挑战。高层模块利用技能图（skill graph），而低层模块采用标准的MARL算法。我们的方法提供了两个贡献。首先，我们在不相关任务的背景下考虑MT-MARL问题，扩展了MTRL的范围。其次，技能图作为标准分层方法的上层，其训练与下层独立进行，能够有效处理不相关任务并增强知识迁移能力。通过大量实验验证了这些优势，并表明所提出的方法优于最新的分层MAPPO算法。视频和代码可在 https://github.com/WindyLab/MT-MARL-SG 获取。

Keyword:

Multi-robot systems

multi-agent reinforcement learning

multi-task reinforcement learning

期刊

IEEE Robotics and Automation Letters

IF:

5.3

论文数:

2.0K

被引数:

3.9W

机构

Beihang University

学者数:

5.2W

论文数: 4.1W

被引数: 37

westlake university

学者数:

5.3K

论文数: 3.7K

被引数: 8

引用论文

Adaptive and Explainable Deployment of Navigation Skills via Hierarchical Deep Reinforcement Learning 基于分层深度强化学习的自适应与可解释导航技能部署

2023 IEEE International Conference on Robotics and Automation (ICRA)

IF0

2023-05-29

OAAI

Kyowoon Lee; Seongun Kim; Jaesik Choi

Swarm Robotics: Past, Present, and Future

PROCEEDINGS OF THE IEEE

IF25.9

2021-07-01

166

OAAI

Dorigo, Marco; Theraulaz, Guy; Trianni, Vito

A Survey on Knowledge Graphs: Representation, Acquisition, and Applications 知识图谱综述: 表示、获取与应用

IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS

IF8.9

2022-02-01

1.3K

OAAI

Ji, Shaoxiong; Pan, Shirui; Cambria, Erik; Marttinen, Pekka; Yu, Philip S.

Knowledge Graph Embedding by Translating on Hyperplanes

Proceedings of the AAAI Conference on Artificial Intelligence

IF0

2014-06-21

OAAI

Zhen Wang; Jianwen Zhang; Jianlin Feng; Zheng Chen

Decentralized Multi-Agent Pursuit Using Deep Reinforcement Learning

IEEE Robotics and Automation Letters

IF0

2021-07-01

OAAI

Cristino de Souza; Rhys Newbury; Akansel Cosgun; Pedro Castillo; Boris Vidolov; Dana Kuli

On Reinforcement Learning for Full-Length Game of StarCraft

Proceedings of the AAAI Conference on Artificial Intelligence

IF0

2019-07-17

OAAI

Zhen-Jia Pang; Ruo-Ze Liu; Zhou-Yu Meng; Yi Zhang; Yang Yu; Tong Lu

Multi-Agent Reinforcement Learning: A Selective Overview of Theories and Algorithms 多智能体强化学习: 理论和算法的选择性综述

Studies in Systems, Decision and Control

IF0

2021-06-24

PREAI

Kaiqing Zhang; Zhuoran Yang; Tamer Başar

A Deep Hierarchical Approach to Lifelong Learning in Minecraft

Proceedings of the AAAI Conference on Artificial Intelligence

IF0

2017-02-12

OAAI

Chen Tessler; Shahar Givony; Tom Zahavy; Daniel Mankowitz; Shie Mannor

Initial Task Allocation in Multi-Human Multi-Robot Teams: An Attention-Enhanced Hierarchical Reinforcement Learning Approach 多人类多机器人团队中的初始任务分配：一种注意力增强的分层强化学习方法

IEEE Robotics and Automation Letters

IF0

2024-04-01

PREAI

Ruiqi Wang; Dezhong Zhao; Arjun Gupte; Byung-Cheol Min

Grandmaster level in StarCraft II using multi-agent reinforcement learning 使用多智能体强化学习在星际争霸II中的大师水平

Nature

IF0

2019-10-30

PREAI

Oriol Vinyals; Igor Babuschkin; Wojciech M. Czarnecki; Michaël Mathieu; Andrew Dudzik; Junyoung Chung; David H. Choi; Richard Powell; Timo Ewalds; Petko Georgiev; Junhyuk Oh; Dan Horgan; Manuel Kroiss; Ivo Danihelka; Aja Huang; Laurent Sifre; Trevor Cai; John P. Agapiou; Max Jaderberg; Alexander S. Vezhnevets; Rémi Leblond; Tobias Pohlen; Valentin Dalibard; David Budden; Yury Sulsky; James Molloy; Tom L. Paine; Caglar Gulcehre; Ziyu Wang; Tobias Pfaff; Yuhuai Wu; Roman Ring; Dani Yogatama; Dario Wünsch; Katrina McKinney; Oliver Smith; Tom Schaul; Timothy Lillicrap; Koray Kavukcuoglu; Demis Hassabis; Chris Apps; David Silver

查看更多内容