EMG-Based Human Motion Analysis: A Novel Approach Using Towel Electrodes and Transfer Learning
EMG-Based Human Motion Analysis: A Novel Approach Using Towel Electrodes and Transfer Learning
复制标题
DOI：
10.1109/jsen.2024.3354307
复制
发表时间：
2024-03
影响因子：
4.3
通讯作者：
Chenyu Tang;Wentian Yi;Sanjeev Kumar;Gurvinder S. Virk;L. Occhipinti
中科院分区：
文献类型：
--
作者：
Chenyu Tang;Wentian Yi;Sanjeev Kumar;Gurvinder S. Virk;L. Occhipinti
这篇文章提出了一种基于肌电(EMG)的人体运动分析系统的创新解决方案，解决了传感器舒适度、个体间差异和劳动密集型标记过程的挑战。该解决方案将基于纺织毛巾的电极与转移学习技术相结合。基于纺织毛巾的石墨烯/PEDOT：PSS复合电极具有生物兼容性、低皮肤阻抗和用户舒适性，而转移学习减少了对大量新数据标记的需要，并增强了运动分析系统的泛化能力。该方法用最少的样本和历元实现了手势的准确分类。这表明了转移学习在有效的基于肌电的人体运动分析中的潜力。
This article presents an innovative solution for electromyography (EMG)-based human motion analysis systems, addressing challenges of sensor comfort, interindividual variations, and labor-intensive labeling processes. The solution combines textile towel-based electrodes with transfer learning techniques. The textile towel-based graphene/PEDOT:PSS composite electrode offers biocompatibility, low skin impedance, and user comfort, while transfer learning reduces the need for extensive new data labeling and enhances the generalization ability of the motion analysis system. The proposed methodology achieves accurate classification of hand gestures with a minimal number of samples and epochs. This demonstrates the potential of transfer learning for efficient EMG-based human motion analysis.