Integral admittance shaping: A unified framework for active exoskeleton control

Abstract

Current strategies for lower-limb exoskeleton control include motion intent estimation, which is subject to inaccuracies in muscle torque estimation as well as modeling error. Approaches that rely on the phases of a uniform gait cycle have proven effective, but lack flexibility to aid other kinds of movement. This research aims at developing a more versatile control that can assist the lower limbs independently of the movement attempted. Our control strategy is based on modifying the dynamic response of the human limbs, specifically their mechanical admittance. Increasing the admittance makes the lower limbs more responsive to any muscle torque generated by the human user.We present Integral Admittance Shaping, a unified mathematical framework for: (a) determining the desired dynamic response of the coupled system formed by the human limb and the exoskeleton, and (b) synthesizing an exoskeleton controller capable of achieving said response.The present control formulation focuses on single degree-of-freedom exoskeleton devices providing performance augmentation. The algorithm generates a desired shape for the frequency response magnitude of the integral admittance (torque-to-angle relationship) of the coupled system. Simultaneously, it generates an optimal feedback controller capable of achieving the desired response while guaranteeing coupled stability and passivity. The potential effects of the exoskeleton’s assistance are motion amplification for the same joint torque, and torque reduction for the same joint motion. The robustness of the derived exoskeleton controllers to parameter uncertainties is analyzed and discussed. Results from initial trials using the controller on an experimental exoskeleton are presented as well.