Extend Single-agent Reinforcement Learning Approach to a Multi-robot Cooperative Task in an Unknown Dynamic Environment Export

@inproceedings{citeulike:1726880, abstract = {Machine learning technology helps multi-robot systems to carry out desired tasks in an unknown dynamic environment. In this paper, we extend the single-agent Q-learning algorithm to a multi-robot box-pushing system in an unknown dynamic environment with random obstacle distribution. There are two kinds of extensions available: directly extending MDP (Markov Decision Process) based Q-learning to the multi-robot domain, and SG-based (Stochastic Game based) Q-learning. Here, we select the first kind of extension because of its simplicity. The learning space, the box dynamics, and the reward function etc. are presented in this paper. Furthermore, a simulation system is developed and its results show effectiveness, robustness and adaptivity of this learning-based multi-robot system. Our statistical analysis of the results also shows that the robots learned correct cooperative strategy even in a dynamic environment.}, author = {Wang, Ying and de Silva, C. W.}, booktitle = {Neural Networks, 2006. IJCNN '06. International Joint Conference on}, citeulike-article-id = {1726880}, citeulike-linkout-0 = {http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=1716795}, journal = {Neural Networks, 2006. IJCNN '06. International Joint Conference on}, keywords = {dynamic\_environment, learning, multi\_agent, reinforcement}, pages = {4999--5005}, posted-at = {2007-10-04 12:09:53}, priority = {4}, title = {Extend Single-agent Reinforcement Learning Approach to a Multi-robot Cooperative Task in an Unknown Dynamic Environment}, url = {http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=1716795}, year = {2006} }

TY - CONF ID - citeulike:1726880 L3 - citeulike-article-id:1726880 N2 - Machine learning technology helps multi-robot systems to carry out desired tasks in an unknown dynamic environment. In this paper, we extend the single-agent Q-learning algorithm to a multi-robot box-pushing system in an unknown dynamic environment with random obstacle distribution. There are two kinds of extensions available: directly extending MDP (Markov Decision Process) based Q-learning to the multi-robot domain, and SG-based (Stochastic Game based) Q-learning. Here, we select the first kind of extension because of its simplicity. The learning space, the box dynamics, and the reward function etc. are presented in this paper. Furthermore, a simulation system is developed and its results show effectiveness, robustness and adaptivity of this learning-based multi-robot system. Our statistical analysis of the results also shows that the robots learned correct cooperative strategy even in a dynamic environment. JF - Neural Networks, 2006. IJCNN '06. International Joint Conference on EP - 5005 TI - Extend Single-agent Reinforcement Learning Approach to a Multi-robot Cooperative Task in an Unknown Dynamic Environment T2 - Neural Networks, 2006. IJCNN '06. International Joint Conference on SP - 4999 KW - dynamic_environment KW - learning KW - multi_agent KW - reinforcement AU - Wang, Ying AU - de Silva, CW PY - 2006/// UR - http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=1716795 ER -