Abstract
In this paper, a multi-agent reinforcement learning method based on action prediction of other agent is proposed. In a multi-agent system, action selection of the learning agent is unavoidably impacted by other agents’ actions. Therefore, joint-state and joint-action are involved in the multi-agent reinforcement learning system. A novel agent action prediction method based on the probabilistic neural network (PNN) is proposed. PNN is used to predict the actions of other agents. Furthermore, the sharing policy mechanism is used to exchange the learning policy of multiple agents, the aim of which is to speed up the learning. Finally, the application of presented method to robot soccer is studied. Through learning, robot players can master the mapping policy from the state information to the action space. Moreover, multiple robots coordination and cooperation are well realized.
Similar content being viewed by others
References
Baghaei KR, Agah A (2007) Multi-agent task allocation for robot soccer. J Intell Syst 16(3): 207–240
Jouffe L (1998) Inference system learning by reinforcement methods. IEEE Trans Syst Man Cybern 28(3): 338–355
Kim JH, Vadakepat P (2000) Multi-agent systems: a survey from the robot-soccer perspective. Int J Intell Autom Soft Comput 6(1): 3–17
Littman ML (1994) Markov games as a framework for multiagent learning. In: Proceedings of the 11th international conference on machine learning. Morgan Kaufmann, San Francisco, pp 157–163
Liu F, Zeng GZ (2006) Multi-agent cooperative learning research based on reinforcement learning. In: 10th international conference on computer supported cooperative work in design. Nanjing, pp 1408–1413
Lu Y, Lin X (2006) Applications of neural network to short-term electric load forecasting. J Shenyang Univ Technol 28(1): 41–44
Moore AW, Atkeson CG (1995) The parti-game algorithm for variable resolution reinforcement learning in multidimensional state-spaces. Mach Learn 21(3): 199–233
Murao H, Kitamura S (1997) Q-learning with adaptive state segmentation (QLASS). In: Proceedings of IEEE international symposium on computational intelligence in robotics and automation. pp 179–184
Reis LP, Lau N, Oliveira EC (2001) Situation based strategic positioning for coordinating a team of homogeneous agents. In: Lecture notes in artificial intelligence, vol 2103. pp 175–197
Specht DF (1990) Probabilistic neural networks. Neural Netw 3(1): 109–118
Specht DF (1990) Probabilistic neural networks and the polynomial adaline as complementary techniques for classification. IEEE Trans Neural Netw 1(1): 111–121
Stone P, Veloso M (2000) Multiagent systems: a survey from a machine learning perspective. Auton Robots 8(3): 345–383
Sutton RS, Barto AG (1998) Reinforcement learning: an introduction. MIT Press Cambridge, Massachusetts
Taniguchi Y, Mori T, Ishii S (2007) Reinforcement learning for cooperative actions in a partially observable multi-agent system. In: Lecture notes in computer science, vol 4668. pp 229–238
Touzet CF (1997) Neural reinforcement learning for behaviour synthesis. Robotics Auton Syst 22(3): 251–281
Veloso M, Stone P, Han K (1998) CMUnited-97: RoboCup-97 small-robot world champion team. AI Magazine 19(3): 61–69
Weinberg M, Rosenschein JS (2004) Best-response multiagent learning in non-stationary environments. In: Proceedings of the thrid international joint conference on autonomous agents and multiagent systems, vol 2. pp 506–513
Wu CJ, Lee TL (2004) A Fuzzy Mechanism for action selection of soccer robots. J Intell Robotic Syst 39(1): 57–70
Xie SM, Chen C, Ding XY (2007) Endpoint prediction of basic-oxygen furnace based on BP neural network. J Shenyang Univ Technol 29(6): 707–710
Yang E, Gu D (2004) Multiagent reinforcement learning for multi-robot systems: a survey. Technical Report CSM-404, Department of Computer Science, University of Essex
Zhong Y (2003) Research on distributed reinforcement learning theory and its application in multi-robot systems. Dissertation for the Degree, Harbin Engineering University
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Duan, Y., Cui, B.X. & Xu, X.H. A multi-agent reinforcement learning approach to robot soccer. Artif Intell Rev 38, 193–211 (2012). https://doi.org/10.1007/s10462-011-9244-8
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10462-011-9244-8