Improvement of Multi-agent Continuous Cooperative Patrolling with Learning of Activity Length

Ayumi Sugiyama, Lingying Wu, Toshiharu Sugawara

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

We propose a learning method that decides the period of activity according to environmental characteristics and the behavioral strategies in the multi-agent continuous cooperative patrol problem. With recent advances in computer and sensor technologies, agents, which are intelligent control programs running on computers and robots, obtain high autonomy so that they can operate in various fields without pre-defined knowledge. However, cooperation/coordination between agents is sophisticated and complicated to implement. We focus on the activity cycle length (ACL) which is the time length from when an agent starts a patrol to when the agent returns to a charging base in the context of a cooperative patrol where agents, like robots, have batteries with limited capacity. A long ACL will enable an agent to visit distant locations, but the agent will require a long rest time to recharge. The basic idea of our method is that if agents have long-life batteries, they can appropriately shorten the ACL, and thus can visit important locations with a short interval of time by recharging frequently. However, appropriate ACL must depend on many elements such as environmental size, number of agents, workload in an environment, and other agents’ behavior and ACLs. Therefore, we propose a method in which agents autonomously learn the appropriate ACL on the basis of the number of events detected per cycle. We experimentally indicate that our agents are able to learn appropriate ACL depending on established spatial divisional cooperation. We also report the details of the analysis of the experimental results to understand the behaviors of agents with different ACLs.

Original languageEnglish
Title of host publicationAgents and Artificial Intelligence - 11th International Conference, ICAART 2019, Revised Selected Papers
EditorsJaap van den Herik, Ana Paula Rocha, Luc Steels
PublisherSpringer
Pages270-292
Number of pages23
ISBN (Print)9783030374938
DOIs
Publication statusPublished - 2019
Event11th International Conference on Agents and Artificial Intelligence, ICAART 2019 - Prague, Czech Republic
Duration: 2019 Feb 192019 Feb 21

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume11978 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference11th International Conference on Agents and Artificial Intelligence, ICAART 2019
CountryCzech Republic
CityPrague
Period19/2/1919/2/21

Keywords

  • Battery limitation
  • Continuous cooperative patrol problem
  • Cycle learning
  • Division of labor
  • Multi-agent

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Fingerprint Dive into the research topics of 'Improvement of Multi-agent Continuous Cooperative Patrolling with Learning of Activity Length'. Together they form a unique fingerprint.

  • Cite this

    Sugiyama, A., Wu, L., & Sugawara, T. (2019). Improvement of Multi-agent Continuous Cooperative Patrolling with Learning of Activity Length. In J. van den Herik, A. P. Rocha, & L. Steels (Eds.), Agents and Artificial Intelligence - 11th International Conference, ICAART 2019, Revised Selected Papers (pp. 270-292). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11978 LNAI). Springer. https://doi.org/10.1007/978-3-030-37494-5_14