Design and Gesture Recognition of Wrist Joint Posture Sensor Based on Fiber Bragg Gratings

• Published in: IEEE sensors journal p. 1
• Main Authors: Yi, Pan, Luo, Binbin, Wu, Decao, Zou, Xue, Chen, Fudan, Huang, Shengyou, Xu, Yiding, Huang, Ling, Shi, Shenghui
• Format: Journal Article
• Language: English
• Published: IEEE 19.10.2024
• Subjects: Deformation; Fiber Bragg grating; Fiber gratings; Forearm muscles; Gesture recognition; LSTM; Muscles; Optical fiber sensors; Optical fibers; Sensors; Skin; Substrates; Temperature sensors; Wrist; Wrist motion
• ISSN: 1530-437X; 1558-1748
• DOI: 10.1109/JSEN.2024.3481013

The wrist joint is the main part of hand movement control, and its posture detection is very important for gesture recognition. This study presents a novel and straightforward method for detecting wrist joint posture utilizing Fiber Bragg Grating (FBG) sensors. FBGs integrated with a composite substrate of Polydimethylsiloxane (PDMS) and Polyimide (PI) were affixed to the skin surface of specific forearm muscles to monitor forearm muscle deformation induced by wrist joint movement. The response characteristics and applicability of the designed sensor were validated through simulation and testing. A Long Short-Term Memory (LSTM) algorithm with an attention mechanism was employed for training and predicting eight static gestures from testers with varying physiques. Experimental results demonstrated that, in comparison to other substrates, FBG with PDMS and PI composite substrates exhibited superior anti-creep properties, fast response and recovery times (198ms/202ms), high sensitivity (7.09pm/°), and good repeatability (Standard deviation of 2.48%). The LSTM model could effectively predict the eight static gestures of the testers, with 99.48% recognition accuracy for a same person and 96.67% recognition accuracy for 10 different persons. The proposed sensors would have potential application in field of the rehabilitation training, virtual reality (VR) or augmented reality (AR), human-computer interaction etc.