Multi-robot task allocation for fire-disaster response based on reinforcement learning

• Published in: 2009 International Conference on Machine Learning and Cybernetics Vol. 4; pp. 2312 - 2317
• Main Authors: Yan-Tao Tian, Mao Yang, Xin-Yue Qi, Yong-Ming Yang
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.07.2009
• Subjects: Convergence; Cybernetics; Distributed control; Distributed task allocation; Dynamic programming; Fires; Learning systems; Machine learning; Multi-robot; Multirobot systems; Reinforcement learning; Robot kinematics; Uncertainty
• ISBN: 9781424437023; 1424437024
• ISSN: 2160-133X
• DOI: 10.1109/ICMLC.2009.5212216

In order to achieve distributed task allocation dynamically and efficiently for multi-robot systems, multi-robot fire-disaster response was presented, because of its dynamic characteristic. The proposed multi-robot task allocation algorithm for fire-disaster response is based on reinforcement learning. The reinforcement learning algorithm for multi-robot is divided into two types: non-cooperation and cooperation, this algorithm satisfies the requirement of dynamic task allocation for fire-disaster response. The experimental results verify that the proposed strategy can achieve efficient multi-robot dynamic task allocation for fire-disaster response, and the fires are extinguished timely.