A reference augmentation design for the adaptive control of a wearable assist robot powered by the McKibben actuator

Abstract

This paper focuses on the control system design for a wearable power assist device which provides force/torque support for people who are doing physically demanding tasks. The device is actuated by the McKibben artificial muscle. The McKibben actuator is know to exhibit high nonlinearity between the supply pressure and the contractile length, and it is basically difficult to design a high performance controller. The present study proposes the use of the Minimal Controller Synthesis algorithm to design a high performance controller for the actuator. The user's joint angular velocity, angular acceleration and the contact force exerted by the user to the frame are fed to the proposed controller to improve the response of the device to the human motion input while providing adequate assist force. Load lifting experiments have been conducted to evaluate the performance of the proposed control system.