Model Predictive Control-Based Assist-as-Needed Strategy for Reducing Motor Slacking in Robot-Assisted Rehabilitation
Sensors · 28.04.2026 · 26 (9) · 2740
DOI: 10.3390/s26092740 ↗Kurz gefasst
Fachartikel in der Zeitschrift Sensors, online erschienen am 28.04.2026. Verfasst von Choonggun Kim und 2 weiteren Autor:innen. Beteiligte Einrichtung: Department of Mechanical Engineering, Sogang University, Seoul 04107, Republic of Korea. Hauptthemen laut MEDLINE: Robotics, Rehabilitation. Die Arbeit nennt 27 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Korea Health Industry Development Institute, Ministry of Trade, Industry and Energy, Ministry of Health & Welfare, Republic of Korea.
Abstract (Originalsprache)
This study proposes a model predictive control (MPC)-based Assist-as-Needed (AAN) strategy for upper-limb rehabilitation robots, with particular emphasis on mitigating motor slacking. In conventional error-based AAN approaches, robotic assistance is regulated through a single coefficient tied to the tracking error; thus, a reduction in voluntary effort is absorbed into the assistive channel and remains obscured by a small tracking error. The proposed method decouples this mechanism by introducing a two-channel admittance structure, in which the robotic-assistance gain Ak and the user-participation-reflection gain Bk are jointly optimized within a single convex MPC formulation. The cost function addresses trajectory tracking, participation-aware force alignment, assistance suppression, and passivity, enforced through energy-tank constraints. The controller was validated in two experiments on a mobile upper-limb rehabilitation robot. The first experiment confirmed differential adaptation of Ak and Bk across three instructed contribution levels, with the participation ratio increasing from 0.103 to 0.879 as the contribution shifted from insufficient to appropriate. The second experiment compared the controller with an error-based AAN baseline and a forgetting-factor AAN baseline under an induced motor-slacking condition, in which the task-direction contribution was reduced to 45%. Under an identical synthesized input, the proposed controller yielded a lower aggregate human-contribution ratio of 0.282, compared with 0.595 and 0.535 for the two baselines, respectively. This indicates that the externally imposed reduction in participation was represented more explicitly in the controller allocation, rather than being masked by error-driven assistive compensation. These results suggest that the proposed approach extends AAN control toward a participation-preserving, anti-slacking strategy for robot-assisted rehabilitation.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Department of Mechanical Engineering, Sogang University, Seoul 04107, Republic of Korea.
- Department of Biomedical Engineering, Asan Medical Center, University of Ulsan College of Medicine, Seoul 05505, Republic of Korea.
ORCID-Profile: Choonggun Kim, Youngjin Moon, Jaesoon Choi
Förderung
- Korea Health Industry Development Institute
- Ministry of Trade, Industry and Energy
- Ministry of Health & Welfare, Republic of Korea
- Ministry of Trade, Industry Industry & Energy
Zitieren
APA
Kim, C., Moon, Y., & Choi, J. (2026). Model Predictive Control-Based Assist-as-Needed Strategy for Reducing Motor Slacking in Robot-Assisted Rehabilitation. Sensors, 26(9), 2740. https://doi.org/10.3390/s26092740
Vancouver
Kim C, Moon Y, Choi J. Model Predictive Control-Based Assist-as-Needed Strategy for Reducing Motor Slacking in Robot-Assisted Rehabilitation. Sensors. 2026;26(9):2740. doi:10.3390/s26092740
BibTeX
@article{Kim2026,
author = {Kim, Choonggun and Moon, Youngjin and Choi, Jaesoon},
title = {Model Predictive Control-Based Assist-as-Needed Strategy for Reducing Motor Slacking in Robot-Assisted Rehabilitation},
journal = {Sensors},
year = {2026},
volume = {26},
number = {9},
pages = {2740},
doi = {10.3390/s26092740}
}
EUVIMED