Finite-Time Model Predictive Force Control for a Cable-Driven Elbow Rehabilitation Exoskeleton Aided by Series Elastic Actuator
Changxian Xu, Shengbo Ma, Baozhen Nie, Zhongbo Sun, Zaixiang Pang
Abstract
Upper limb rehabilitation exoskeletons demonstrate application potential as intelligent medical devices for motor function reconstruction for stroke patients. However, the existing upper limb devices are difficult to meet the demands of clinical applications due to their bulky design and poor flexibility. This letter presents a lightweight elbow rehabilitation exoskeleton (ERE) with compliant properties actuated by a Bowden cable-driven mechanism. A series elastic actuator (SEA) with dual compliant units is designed to deliver precise torque to robot joints while ensuring real-time safety for the subjects. Furthermore, a model predictive force control (MPFC) scheme is proposed for the cable-driven ERE with SEA used to provide precise torque control for subjects. Meanwhile, a finite-time convergent neural dynamics (FTCND) model is constructed to solve the optimal solution of the MPFC scheme. Theoretical analyses, numerical simulations, and physical experiments verify that the proposed MPFC scheme solved by the FTCND has the advantages of finite-time convergence, strong stability, and robustness. Evaluation with the K5 metabolic acquisition device indicates that wearing the exoskeleton under the proposed MPFC scheme significantly reduces the respiratory metabolic level of the subjects, with reductions of approximately 13.75%–15.90% in O$_{2}$ uptake and 5.07%–28.86% in CO$_{2}$ output.
BibTeX
@inproceedings{ral2026_finitetimemodelp,
title = {Finite-Time Model Predictive Force Control for a Cable-Driven Elbow Rehabilitation Exoskeleton Aided by Series Elastic Actuator},
author = {Changxian Xu and Shengbo Ma and Baozhen Nie and Zhongbo Sun and Zaixiang Pang},
booktitle = {RA-L 2026},
year = {2026}
}