European Journal Of Advances In Engineering And Technology
Published December 31, 2022 | Version v1
Journal article Open

Comparison between a centralized and decentralized method for multi-agent collaboration in an exploration problem

Authors/Creators

  • 1. Laboratory of Computer Science Engineering and Automation, University of Douala, P. O. Box 1872 Douala, Cameroon

Description

ABSTRACT

In this paper, two methods for robot collaboration in a multi-agent system (MAS) exploration problem based on reinforcement learning are presented. In this problem, the agents are placed in an arena where a target is located and the goal is to measure the time taken by the robots to detect and destroy it. The experiment was carried out several times in the Robotarium's Matlab API (Application Programming Interface) in order to compare the times taken by each method. In the first method, called centralised, the agents move in groups and traverse the arena from one end to the other, whereas in the second, called decentralised, each agent moves autonomously. The experimental results show that both approaches, distributed and centralized, can finally solve the problem, but the coordination performance of the proposed centralized approach is much better than that of the decentralized approach. We demonstrate through all these experiments that for exploration in an arena with target localization, centralized methods are more efficient because they take less time than decentralized methods.

Files

EJAET-9-12-21-31.pdf

Files (396.1 kB)

Name Size Download all
md5:06648614f8474cd4557bd4a1e729b9b7
396.1 kB Preview Download

Additional details

References

  • [1]. C. R. Kube and E. Bonabeau, "Cooperative transport by ants and robots," Robotics and Autonomous Systems, vol. 30, no 1-2, pp. 85-101, Jan. 2000, doi: 10.1016/S0921-8890(99)00066-4.
  • [2]. J. Nembrini and A. Martinoli, "Robotics in Swarms, Results and Future Directions", p. 6.
  • [3]. G. Vásárhelyi, C. Virágh, G. Somorjai, T. Nepusz, A. E. Eiben, and T. Vicsek, "Optimized flocking of autonomous drones in confined environments," Sci. Robot. , vol. 3, no 20, p. eaat3536, Jul 2018, doi: 10.1126/scirobotics.aat3536.
  • [4]. D. Pickem et al , "The Robotarium: A remotely accessible swarm robotics research testbed," in 2017 IEEE International Conference on Robotics and Automation (ICRA), May 2017, pp. 1699-1706. doi: 10.1109/ICRA.2017.7989200.
  • [5]. D. Pickem et al , "Safe, Remote-Access Swarm Robotics Research on the Robotarium". arXiv, 3 April 2016. Accessed: 24 August 2022. [Online]. Available at: http://arxiv.org/abs/1604.00640
  • [6]. M. Rubenstein, A. Cornejo, and R. Nagpal, "Programmable self-assembly in a thousand-robot swarm," Science, vol. 345, no 6198, pp. 795-799, August 2014, doi: 10.1126/science.1254295.
  • [7]. M. Dorigo, G. Theraulaz, and V. Trianni, "Reflections on the future of swarm robotics," Sci. Robot. , vol. 5, no 49, p. eabe4385, Dec. 2020, doi: 10.1126/scirobotics.abe4385.
  • [8]. S. Mitri, S. Wischmann, D. Floreano, and L. Keller, "Using robots to understand social behaviour", Biol Rev Camb Philos Soc, vol. 88, no 1, pp. 31-39, Feb 2013, doi: 10.1111/j.1469-185X.2012.00236.x.
  • [9]. K. N. McGuire, C. De Wagter, K. Tuyls, H. J. Kappen, and G. C. H. E. de Croon, "Minimal navigation solution for a swarm of tiny flying robots to explore an unknown environment," Science Robotics, vol. 4, no 35, p. eaaw9710, Oct 2019, doi: 10.1126/scirobotics.aaw9710.
  • [10]. J. Krause, A. F. T. Winfield, and J.-L. Deneubourg, "Interactive robots in experimental biology," Trends Ecol Evol, vol. 26, no 7, pp. 369-375, Jul 2011, doi: 10.1016/j.tree.2011.03.015.
  • [11]. L. Busoniu, R. Babuska, and B. De Schutter, "A Comprehensive Survey of Multiagent Reinforcement Learning", IEEE Trans. Syst., Man, Cybern. C, vol. 38, no 2, pp. 156-172, March 2008, doi: 10.1109/TSMCC.2007.913919.
  • [12]. C. Yu, Y. Dong, Y. Li, and Y. Chen, "Distributed multi-agent deep reinforcement learning for cooperative multi-robot pursuit," J. eng. , vol. 2020, no 13, pp. 499-504, Jul. 2020, doi: 10.1049/joe.2019.1200.
  • [13]. A. T. Hayes and P. Dormiani-Tabatabaei, "Self-organized flocking with agent failure: Off-line optimization and demonstration with real robots," in Proceedings 2002 IEEE International Conference on Robotics and Automation (Cat. No.02CH37292), Washington, DC, USA, 2002, vol. 4, pp. 3900-3905. doi: 10.1109/ROBOT.2002.1014331.
  • [14]. R. Zlot and A. Stentz, "Market-based Multirobot Coordination for Complex Tasks," The International Journal of Robotics Research, vol. 25, no 1, pp. 73-101, Jan. 2006, doi: 10.1177/0278364906061160.
  • [15]. R. S. Sutton and A. G. Barto, "Reinforcement Learning: An Introduction", p. 551.
  • [16]. "Artificial Intelligence A-ZTM: Learn How To Build An AI", Udemy. https://www.udemy.com/course/artificial-intelligence-az/ (accessed 18 September 2022).
  • [17]. M. van Otterlo, "Markov Decision Processes: Concepts and Algorithms", p. 23.
  • [18]. D. J. White, "A Survey of Applications of Markov Decision Processes", vol. 44, no 11, p. 25.
  • [19]. H. van Hasselt, A. Guez, and D. Silver, "Deep Reinforcement Learning with Double Q-Learning", p. 7.
  • [20]. A. Juliani, "Simple Reinforcement Learning with Tensorflow Part 0: Q-Learning with Tables and Neural Networks", Emergent // Future, 26 May 2017. https://medium.com/emergent-future/simple-reinforcement-learning-with-tensorflow-part-0-q-learning-with-tables-and-neural-networks-d195264329d0 (accessed 18 September 2022).