Implementation of LPV H∞ Loop-Shaping Control for a Variable Stiffness Actuator

Abstract

Abstract Compliant actuators have been increasingly used for active joints in lower-limb exoskeletons or orthoses because they help to guarantee a safe human interaction. One example of such compliant motors is the variable stiffness actuator (VSA). The design of a torque controller for such an actuator is a crucial task in order to provide patients with physical gait assistance and overcome the mechanical limitations of the VSA. Our goal is to implement a torque controller for our mechanical-rotary variable impedance actuator (MeRIA) used in future lower-limb exoskeletons. In the torque control design, we derive a gain-scheduled controller for the polytopic linear parameter-varying (LPV) model of the actuator. This controller is based on the classical H∞ loop-shaping approach. Measurements on the hardware-in-the-loop system in time and frequency domain show that the designed controller provides adequate performance over the whole varying stiffness range. Additionally, the controller provides H∞ robustness with respect to coprime factor uncertainty for the polytopic system. Thus, the torque controller fulfills major safety requirements, and can further be used for human-in-the-loop tests and applications with a lower-limb exoskeleton.