Graphical Minimax Game and On-Policy Reinforcement Learning for Consensus of Leaderless Multi-Agent Systems *

Wei Dong; Chunyan Wang; Jinna Li; Jianan Wang

doi:10.1109/ICCA51439.2020.9264527

Graphical Minimax Game and On-Policy Reinforcement Learning for Consensus of Leaderless Multi-Agent Systems *

Wei Dong, Chunyan Wang, Jinna Li, Jianan Wang

School of Aerospace Engineering

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

1 Citation (Scopus)

Abstract

In this paper, we study the adaptive optimal consensus control of leaderless multi-agent systems (MASs) with heterogeneous dynamics. First, the consensus control problem is converted into a graphical minimax game problem and the corresponding algebraic Riccati equation (ARE) for each agent is obtained. Then, an on-policy reinforcement learning algorithm is proposed to online learn the optimal control policy without requiring the system dynamics. A certain rank condition is established to guarantee the convergence of the proposed online learning algorithm to the unique solution of the ARE. Finally, the effectiveness of the proposed algorithm is demonstrated through a numerical simulation.

Original language	English
Title of host publication	2020 IEEE 16th International Conference on Control and Automation, ICCA 2020
Publisher	IEEE Computer Society
Pages	606-611
Number of pages	6
ISBN (Electronic)	9781728190938
DOIs	https://doi.org/10.1109/ICCA51439.2020.9264527
Publication status	Published - 9 Oct 2020
Event	16th IEEE International Conference on Control and Automation, ICCA 2020 - Virtual, Sapporo, Hokkaido, Japan Duration: 9 Oct 2020 → 11 Oct 2020

Publication series

Name	IEEE International Conference on Control and Automation, ICCA
Volume	2020-October
ISSN (Print)	1948-3449
ISSN (Electronic)	1948-3457

Conference

Conference	16th IEEE International Conference on Control and Automation, ICCA 2020
Country/Territory	Japan
City	Virtual, Sapporo, Hokkaido
Period	9/10/20 → 11/10/20

Access to Document

10.1109/ICCA51439.2020.9264527

Cite this

Dong, W., Wang, C., Li, J., & Wang, J. (2020). Graphical Minimax Game and On-Policy Reinforcement Learning for Consensus of Leaderless Multi-Agent Systems *. In 2020 IEEE 16th International Conference on Control and Automation, ICCA 2020 (pp. 606-611). Article 9264527 (IEEE International Conference on Control and Automation, ICCA; Vol. 2020-October). IEEE Computer Society. https://doi.org/10.1109/ICCA51439.2020.9264527

@inproceedings{571dfc04bc6a40d4a17dd2023dd38ff5,

title = "Graphical Minimax Game and On-Policy Reinforcement Learning for Consensus of Leaderless Multi-Agent Systems *",

abstract = "In this paper, we study the adaptive optimal consensus control of leaderless multi-agent systems (MASs) with heterogeneous dynamics. First, the consensus control problem is converted into a graphical minimax game problem and the corresponding algebraic Riccati equation (ARE) for each agent is obtained. Then, an on-policy reinforcement learning algorithm is proposed to online learn the optimal control policy without requiring the system dynamics. A certain rank condition is established to guarantee the convergence of the proposed online learning algorithm to the unique solution of the ARE. Finally, the effectiveness of the proposed algorithm is demonstrated through a numerical simulation.",

author = "Wei Dong and Chunyan Wang and Jinna Li and Jianan Wang",

note = "Publisher Copyright: {\textcopyright} 2020 IEEE.; 16th IEEE International Conference on Control and Automation, ICCA 2020 ; Conference date: 09-10-2020 Through 11-10-2020",

year = "2020",

month = oct,

day = "9",

doi = "10.1109/ICCA51439.2020.9264527",

language = "English",

series = "IEEE International Conference on Control and Automation, ICCA",

publisher = "IEEE Computer Society",

pages = "606--611",

booktitle = "2020 IEEE 16th International Conference on Control and Automation, ICCA 2020",

address = "United States",

}

Dong, W, Wang, C, Li, J & Wang, J 2020, Graphical Minimax Game and On-Policy Reinforcement Learning for Consensus of Leaderless Multi-Agent Systems *. in 2020 IEEE 16th International Conference on Control and Automation, ICCA 2020., 9264527, IEEE International Conference on Control and Automation, ICCA, vol. 2020-October, IEEE Computer Society, pp. 606-611, 16th IEEE International Conference on Control and Automation, ICCA 2020, Virtual, Sapporo, Hokkaido, Japan, 9/10/20. https://doi.org/10.1109/ICCA51439.2020.9264527

Graphical Minimax Game and On-Policy Reinforcement Learning for Consensus of Leaderless Multi-Agent Systems *. / Dong, Wei; Wang, Chunyan; Li, Jinna et al.
2020 IEEE 16th International Conference on Control and Automation, ICCA 2020. IEEE Computer Society, 2020. p. 606-611 9264527 (IEEE International Conference on Control and Automation, ICCA; Vol. 2020-October).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Graphical Minimax Game and On-Policy Reinforcement Learning for Consensus of Leaderless Multi-Agent Systems *

AU - Dong, Wei

AU - Wang, Chunyan

AU - Li, Jinna

AU - Wang, Jianan

PY - 2020/10/9

Y1 - 2020/10/9

N2 - In this paper, we study the adaptive optimal consensus control of leaderless multi-agent systems (MASs) with heterogeneous dynamics. First, the consensus control problem is converted into a graphical minimax game problem and the corresponding algebraic Riccati equation (ARE) for each agent is obtained. Then, an on-policy reinforcement learning algorithm is proposed to online learn the optimal control policy without requiring the system dynamics. A certain rank condition is established to guarantee the convergence of the proposed online learning algorithm to the unique solution of the ARE. Finally, the effectiveness of the proposed algorithm is demonstrated through a numerical simulation.

AB - In this paper, we study the adaptive optimal consensus control of leaderless multi-agent systems (MASs) with heterogeneous dynamics. First, the consensus control problem is converted into a graphical minimax game problem and the corresponding algebraic Riccati equation (ARE) for each agent is obtained. Then, an on-policy reinforcement learning algorithm is proposed to online learn the optimal control policy without requiring the system dynamics. A certain rank condition is established to guarantee the convergence of the proposed online learning algorithm to the unique solution of the ARE. Finally, the effectiveness of the proposed algorithm is demonstrated through a numerical simulation.

UR - http://www.scopus.com/inward/record.url?scp=85098075592&partnerID=8YFLogxK

U2 - 10.1109/ICCA51439.2020.9264527

DO - 10.1109/ICCA51439.2020.9264527

M3 - Conference contribution

AN - SCOPUS:85098075592

T3 - IEEE International Conference on Control and Automation, ICCA

SP - 606

EP - 611

BT - 2020 IEEE 16th International Conference on Control and Automation, ICCA 2020

PB - IEEE Computer Society

T2 - 16th IEEE International Conference on Control and Automation, ICCA 2020

Y2 - 9 October 2020 through 11 October 2020

ER -

Dong W, Wang C, Li J, Wang J. Graphical Minimax Game and On-Policy Reinforcement Learning for Consensus of Leaderless Multi-Agent Systems *. In 2020 IEEE 16th International Conference on Control and Automation, ICCA 2020. IEEE Computer Society. 2020. p. 606-611. 9264527. (IEEE International Conference on Control and Automation, ICCA). doi: 10.1109/ICCA51439.2020.9264527

Graphical Minimax Game and On-Policy Reinforcement Learning for Consensus of Leaderless Multi-Agent Systems *

Abstract

Publication series

Conference

Access to Document

Other files and links

Fingerprint

Cite this