A model of emotional intelligent agent for cooperative goal exploration

Authors:
Takashi Kuremoto;Tetsuya Tsurusaki;Kunikazu Kobayashi;Shingo Mabu;Masanao Obayashi
Affiliations:
Graduate School of Science and Engineering, Yamaguchi University, Ube, Yamaguchi, Japan;Graduate School of Science and Engineering, Yamaguchi University, Ube, Yamaguchi, Japan;School of Information Science & Technology, Aichi Prefectural University, Nagakute, Aichi, Japan;Graduate School of Science and Engineering, Yamaguchi University, Ube, Yamaguchi, Japan;Graduate School of Science and Engineering, Yamaguchi University, Ube, Yamaguchi, Japan
Venue:
ICIC'13 Proceedings of the 9th international conference on Intelligent Computing Theories
Year:
2013

Citing 6
Cited 0

Cooperative behavior acquisition for mobile robots in dynamically changing real worlds via vision-based reinforcement learning and development

Artificial Intelligence - Special issue on Robocop: the first step
Introduction to Reinforcement Learning

Introduction to Reinforcement Learning
Cooperative Mobile Robotics: Antecedents and Directions

Autonomous Robots
Metalearning and neuromodulation

Neural Networks - Computational models of neuromodulation
Autonomic behaviors of swarm robots driven by emotion and curiosity

LSMS/ICSEE'10 Proceedings of the 2010 international conference on Life system modeling and simulation and intelligent computing, and 2010 international conference on Intelligent computing for sustainable energy and environment: Part III
Cooperative behavior acquisition in multi-agent reinforcement learning system using attention degree

ICONIP'12 Proceedings of the 19th international conference on Neural Information Processing - Volume Part III

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this study, a novel model of the intelligent agent is proposed by introducing a dynamic emotion model into conventional action selection policy of the reinforcement learning method. Comparing with the conventional Q-learning of reinforcement learning, the proposed method adds two emotional factors in to the state-action value function: "arousal value" factor which affects motivation of action and "pleasure value" factor which influences the probability of action selection. The emotional factors are affected by the other agents when multiple agents exist in the perception area. Computer simulations of pursuit problems of static/dynamic preys were performed and all results showed effectiveness of the proposed method, i.e., faster learning convergence was confirmed comparing with the case of conventional Q-learning method.