A Biologically Inspired Adaptive Model Theory For Humanoid Robot Arm Control

Abstract

Biological control systems (which deal with non-stiff-joint plants, as a human arm is) have evolved during millions of years and have become into an interesting paradigm to emulate in robotic controller construction. The cerebellum is known to be involved in control and learning of smooth coordinated movements. Furthermore, an accurate understanding of how this advance control engine works should have a strong impact in controlling biomorphic robots. We propose a decentralized motor learning model for the cerebellum, as well as an intelligent adaptive system based on historical physiological data. The humanoid model is used to create a hybrid force/position controller for a dual arm.