A Sliding Mode Controller Using an LS-SVM Model for a Water-Hydraulic Artificial Rubber Muscle

Abstract

We describe a sliding mode controller design for an artificial rubber muscle driven by tap-water pressure. The hysteretic characteristics of this water-hydraulic artificial rubber muscle (WARM) often deteriorate its control accuracy. To cope with this complicated hysteresis, a modeling approach based on the least squares support vector machine (LS-SVM) with nonlinear kernel functions is first applied to a WARM. By employing this LS-SVM-based WARM model, a sliding mode controller is then derived for the WARM drive system. We verify the control performance of the proposed controller and compare its tracking accuracy with our previously developed controller through experiments.