Autonomous control of real snake-like robot using reinforcement learning; Abstraction of state-action space using properties of real world

摘要

In this paper we consider autonomous control of a real snake-like robot using reinforcement learning. We focus on curse of dimensionality and lack of generality, and point out that the causes of the problems are not in learning algorithm but in neglect of properties of the real world. To solve the problems we propose new framework in which body of robot abstract general meaning by using properties of the real world as information processor. We apply the proposed framework for controlling a snake-like robot and confirm that the two problems are solved simultaneously, without changing learning algorithm at all. To demonstrate the effectiveness of the proposed framework experiments has been carried out.