1A1-D10 Cooperative Behavior Acquisition with Reinforcement Learning Robots Based on the Mechanism of Selecting the State Space Representations(Evolution and Learning for Robotics(1))
Koki Kage, Junki Sakanoue, Toshiyuki Yasuda, Kazuhiro Ohkura · The Proceedings of JSME annual Conference on Robotics and Mechatronics (Robomec) · 2012
Multi-robot systems (MRS) can be expected to solve a task which one robot cannot perform. In MRS, reinforcement learning is one of promising approaches for controlling each robot. However, its performance depends a great deal on the segmentation of state and action spaces. To deal with this problem, we have been developing a new technique which segments state and action spaces autonomously, named BRL. In order to improve the learning performance, this paper introduces mechanism of selecting either of two state spaces: one is parametric model useful for exploration and the other is non-parametric model for exploitation. We investigate our proposed technique, by conducting physical experiments for a cooperative carrying task with three autonomous mobile robots.