This paper proposes learning method in dynamicenvironment using Hierarchical modular reinforcement learning.Hierarchical modular reinforcement learning was prorosed byWatanabe, consists of 2 layered learning where Profit-Sharingworks to plan a target position in higher layer and Q-learningtrains the state-action pair to the target in lower layer. It candivide task and state space, then reduce state dimension andimprove learning capability. In this paper, we analyzed multiagentbehavior the pursuit problem in dynamic environment.
展开▼
