An Electromyographic‐Based Control Using Gaussian Mixture Model on an Upper‐Limb Cable‐Driven Rehabilitation Robot
Electromyographic (EMG)‐based admittance control by arm force can provide continuous motion control in robot‐assisted rehabilitation. Natural and complex physical human–robot interactions utilizing intelligent EMG‐based interfaces require a computational estimation model for 3D voluntary forces. Exi...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Wiley
2025-04-01
|
| Series: | Advanced Intelligent Systems |
| Subjects: | |
| Online Access: | https://doi.org/10.1002/aisy.202400505 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Electromyographic (EMG)‐based admittance control by arm force can provide continuous motion control in robot‐assisted rehabilitation. Natural and complex physical human–robot interactions utilizing intelligent EMG‐based interfaces require a computational estimation model for 3D voluntary forces. Existing computational models infrequently directly encode the interconnections among the spatial‐dimensional components of 3D arm forces and experience performance degradation caused by confounding variables like muscle fatigue in EMG‐based control. Motivated by this challenge, a computational model is proposed using the Gaussian mixture model (GMM), which only requires a user‐friendly calibration by an incremental GMM approach to reduce the effect of muscle fatigue. GMM aims to encode the multivariable connection between EMG and voluntary force by joint probability density distribution. The EMG‐based control system retrieves the estimated voluntary forces given EMG from GMM by utilizing the Gaussian mixture regression. Herein, the performance of the proposed EMG‐based admittance control is tested using GMM by trajectory‐tracking experiments before and after the fatigue‐inducing experiment. The experiments include eight healthy participants. The experimental outcomes prove that the EMG‐based control using calibrated GMM shows an increase of 23.66% and 8.17% in tracking precision and motion compliance, showing potential application across diverse domains involving physical human–robot interactions. |
|---|---|
| ISSN: | 2640-4567 |