Hierarchical intelligent control of modular manipulators Part B: Reconfigurability and experimental validation

Abstract

For pt.A, see ibid., p.2-7 (2003). In part A of this paper, we developed an intelligent neurofuzzy architecture that can be easily used in the presence of dynamic parameter uncertainty and unmodeled disturbances to control modular and reconfigurable manipulators. The proposed architecture has several levels of hierarchy built on top of a conventional PID controller. The present part B of the paper discussed systematic guidelines to design the skill module of the neurofuzzy control. Such module is used to update the adaptive control parameters of the neurofuzzy architecture when the robotic arm is reconfigured. Furthermore, in this part B of the paper, we present experiments that where conducted on a modular and reconfigurable robot. Some of the most notably significant experimental results are reported.