Hand Gesture Recognition is Achieved by Collecting the Deformation of Arm Muscles Using Cladding Waveguide Fiber Bragg Gratings

In the field of medical rehabilitation, smart wearable technology is highly favored for its features such as portability, flexibility, and comfort. This paper proposes a method that uses Cladding Waveguide Fiber Bragg Grating (CLFBG) sensors to identify five static hand gestures based on forearm muscle deformation. Compared with traditional Fiber Bragg Grating (FBG), CLFBG demonstrates a larger wavelength shift (109 pm higher) and faster response time under identical experimental conditions. This method encapsulates CLFBG with Polydimethylsiloxane (PDMS). CLFBG are fabricated through micromachining using a femtosecond laser. The sensors are attached to the skin surface above specific forearm muscles to monitor the deformation of the forearm muscles caused by different static hand gestures. The deformation of the forearm muscles causes a shift in the central wavelength of the CLFBG. By combining the central wavelength shifts of the CLFBG caused by different hand movements with a Multi Layer Perceptron (MLP) + SHapley Additive exPlanations (SHAP) model, it is very effective to predict five hand movements of the test subjects. The training accuracy is 99.19%, and the prediction accuracy is 98.26% At the same time, the influence factors of the three CLFBGs in the training and prediction model are obtained. The proposed sensor provides an effective application solution for rehabilitation training, human posture analysis, and virtual reality.