Coordination in multiagent reinforcement learning systems

M. A.S. Kamal, Junichi Murata

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

This paper presents a novel method for real-time coordination control of multiagent systems in maximizing global benefits keeping a balance with individual benefits of agents. In this coordination mechanism a reinforcement-learning agent learns to select its action estimating global state value and immediate reward. The estimated global state value of the system makes an agent cooperative with others. This learning method is implemented in the test bed multiagent transportation service control for a city. The outstanding performance of the proposed. method in different aspects compared to other heuristic methods indicates its effectiveness for multiagent cooperative systems.

Original languageEnglish
Pages (from-to)1197-1204
Number of pages8
JournalLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume3213
Publication statusPublished - Dec 1 2004

    Fingerprint

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

Cite this