Juggling control using neural oscillators

In this paper we discuss pattern generations and switching for robot manipulators to perform tasks under qualitative changes of the dynamic environment. The approach is to compose continuous type artificial neural elements as an oscillator based on mutual entrainment and generate trajectories for the manipulators. Using the neural oscillator, juggling control of manipulators is considered as a concrete task, and a method to alternate patterns of the trajectories according to the change of the desired value and environment is proposed. Some numerical examples show the efficiency of the proposed method.<
>