Robot arm control using entrainment property of central pattern generator

Abstract

Robot arm control based on a neural oscillator is addressed to achieve biologically inspired robot motion. It is known that neural oscillator networks could produce rhythmic commands efficiently and robustly under the changing task environment. The entrainment feature of neural oscillators enables a robot to exhibit natural adaptive motions as a controller. For technically attaining this, we use a two-link robot arm whose joints are connected to the most widely used Matsuoka's neural oscillator. Then we present an approach to perform various given tasks exploiting the neural oscillator and its entrainment property. In addition, we experimentally demonstrate an impressive capability of self-adaptation of the neural oscillator that enables the robotic arm to make adaptive changes from a given task into an appropriate motion corresponding to outer conditions.