Motor Training Using Mental Workload (MWL) With an Assistive Soft Exoskeleton System: A Functional Near-Infrared Spectroscopy (fNIRS) Study for Brain–Machine Interface (BMI)

Asgher, Umer; Khan, Muhammad Jawad; Asif Nizami, Muhammad Hamza; Khalil, Khurram; Ahmad, Riaz; Ayaz, Yasar; Naseer, Noman · 2021 · Crossref

DOI: 10.3389/fnbot.2021.605751

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study addresses the need for accessible, neuroergonomic assistive technologies for patients with motor disabilities, such as those caused by stroke or tetraplegia. The authors propose a Brain–Machine Interface (BMI) system that utilizes Mental Workload (MWL) signals to control a lightweight, soft robotic exoskeleton hand. The motivation stems from the limitations of existing BMI systems, which often require complex protocols or deliberate muscle control that may be difficult for patients under stress or cognitive fatigue. By leveraging functional near-infrared spectroscopy (fNIRS), the system aims to provide a non-invasive, portable interface that translates cognitive states into motor commands, thereby enhancing patient autonomy and ease of use. The experimental design involved 15 healthy participants who performed hand-grasping tasks while wearing a 12-channel portable fNIRS system positioned over the pre-frontal cortex (PFC). The system recorded hemodynamic changes in oxygenated (HbO) and deoxygenated (HbR) hemoglobin at an 8 Hz sampling rate. Participants engaged in two levels of mental arithmetic tasks to induce distinct MWL states: Level 1 involved simple three-digit addition, while Level 2 required more complex operations involving short-term memory and sequential calculations. Task difficulty was validated using the NASA Task Load Index (TLX). The acquired fNIRS signals were processed to extract mean and slope features, which were then classified using a Support Vector Machine (SVM) algorithm. These classified signals generated binary commands ("open" and "close") to operate a custom-designed, five-degree-of-freedom soft exoskeleton hand. The exoskeleton features servo-tendon actuation for independent finger control, optimized for smooth trajectories and effective force transmission during grasping. The results demonstrated the feasibility of the proposed BMI system. The SVM classifier achieved a maximum classification accuracy of 91.31% when using a combination of mean and slope features, with an average accuracy of 87.9%. The system yielded an average information transfer rate (ITR) of 1.43. The NASA-TLX scores confirmed that Level 2 tasks induced significantly higher mental workload than Level 1, validating the distinctiveness of the cognitive states used for classification. The integration of the fNIRS-based classifier with the soft exoskeleton allowed for reliable control of the robotic hand based on cognitive load rather than explicit motor imagery or muscle activation. The significance of this work lies in its demonstration of a passive BMI approach using MWL for assistive robotics. By relying on cognitive load rather than precise motor control, the system offers a robust alternative for patients who may struggle with traditional BCI paradigms. The study highlights the potential of fNIRS-based neuroergonomic interfaces to simplify human-machine interaction, providing a viable pathway for developing lightweight, user-friendly assistive devices for hemiplegic patients. This approach contributes to the field of neurorehabilitation by emphasizing ergonomic design and cognitive accessibility in the development of next-generation prosthetic technologies.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-nvidia on 2026-08-10; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• neuro workload indices