Denoising Textile Kinematics Sensors: A Machine Learning Approach
Denoising Textile Kinematics Sensors: A Machine Learning Approach
复制标题
DOI：
--
复制
发表时间：
--
期刊：
影响因子：
--
通讯作者：
Yuxuan Han;Vigyanshu Mishra;A. Kiourti
中科院分区：
文献类型：
--
作者：
Yuxuan Han;Vigyanshu Mishra;A. Kiourti
监测现实环境中人体的运动学有利于医疗保健、体育、人机界面等多种应用。为此，我们最近报道了新型可穿戴式基于纺织品的传感器，它们由发射/接收环路组成，并根据法拉第定律进行操作，以无缝监测关节弯曲角度（例如膝盖、肘部等）。然而，一旦嵌入织物中，环路就会随着织物的运动而漂移，从而损害传感器的运行。在这项工作中，我们报告了一种机器学习方法，用于建模和消除与织物上变形的电子纺织品传感器相关的噪声（即电子纺织品噪声），重点是运动学监控应用。
Monitoring kinematics of the human body in real-world environments is beneficial to applications as diverse as healthcare, sports, human-machine interfaces, and more. To this end, we recently reported new classes of wearable textile-based sensors that consist of transmit/receive loops and operate based on Faraday’s Law to seamlessly monitor joint flexion angles (e.g., knee, elbow, etc.). However, once embedded in fabrics, the loops will drift along with fabric movement and hence will impair the sensor’s operation. In this work, we report a machine learning approach to model and remove noise associated with e-textile sensors being deformed upon the fabric (namely, e-textile noise), with a focus on kinematics monitoring applications.