A Machine Learning-Assisted Self-Powered and Wearable Optical Sensor Glove for Gesture Recognition

作者：Yang Zou, Chenxu Lu, Chaomao Zhang, Guozhen Chen, Renfei Kuang, Zhenshi Chen, Rui Min, Quandong Huang, Shuyan Zhu, Yungen Peng, Jingcui Song, Qingming Chen · 发表于：Journal of Lightwave Technology · 年份：2025 · DOI：10.1109/jlt.2025.3572078 · 被引用次数：5 · 研究领域：Hand Gesture Recognition Systems

Gesture recognition is a vital component of human-computer interaction (HCI), which plays a key role in artificial intelligent (AI) technology. Self-powered, portable and wearable sensors are highly desirable for HCI applications. Optical wearable sensors have gained widespread attention in gesture recognition due to their high sensitivity and great biocompatibility. This paper reports a novel mechanoluminescent (ML) polymer optical fiber-based glove designed for self-powered and portable gesture recognition. The elastic ML optical fiber directly converts finger movements into fluorescence for strain sensing. Combined with the supplementary alumina nanoparticles, the fluorescence intensity of ZnS: Cu ML phosphors has been enhanced by 7 times to achieve a highperformance strain sensor. By integrating five ML fibers into a flexible printed circuit (FPC) with a wireless communication unit, we developed a smart glove capable of gesture detection. With the assistance of machine-learning training, the proposed glove can accurately identify six hand gestures and five grasping actions with recognition accuracies of 100% and 90%, respectively. This portable intelligent glove operates without external light sources and electrical drivers, making it highly suitable for potential applications in HCI and AI fields