Using a Wearable Wristband Integrated with a Self-Supervised Contrastive Learning Algorithm

- Paper published in the journal Nano-Micro Letters (Publisher: Springer), 2023 JCR Q1, IF=31.6 (Vol. 17, Article No. 41, 2024.10.16) -

Professor Jongcheol Lee's research team from the Department of Electronics Convergence Engineering at Kwangwoon University, in international collaboration with Professor Yang Li’s team from Shangdong University in Zhen'an, China, Professor Guojian Xun’s team from Beijing Institute of Technology (BIT), and Professor Namyoung Kim’s team from Kwangwoon University, has developed a rapid application method for dynamic air-writing recognition using a wearable wristband integrated with a self-supervised contrastive learning algorithm.

In this study, a self-supervised contrastive AI algorithm was applied to a wearable wristband equipped with a hydrogel pressure sensor to enable real-time recognition of air-writing. The utility of this approach was demonstrated through high-accuracy recognition of numbers and letters and allowing input without a keyboard with a high level of reliability. Such a wearable wristband has significant impact, as it integrates seamlessly into users' daily lives and provides an intuitive means of communicating with digital devices through gestures without interruption.

This research was supported by the 2023 University Academic Research Fund and the BK21 Phase 4 project. It was published in the journal Nano-Micro Letters (Q1, IF=31.6, JIF=98.2) under the title “A Rapid Adaptation Approach for Dynamic Air-Writing Recognition Using Wearable Wristbands with Self-Supervised Contrastive Learning.”

The journal link is as follows. https://doi.org/10.1007/s40820-024-01545-8

a. A wearable wristband for air-writing recognition, equipped with four hydrogel pressure sensors (D1-D4) combined with a wireless Wi-Fi module

b. (i) An iontronic device composed of AgNWs/PVA electrodes and a micro-cone structured photocurable ionic hydrogel, (ii) a simplified block diagram of the wristband system and a customized user interface

c. Real-time prediction and display of air-writing through TS-VFC learning. Rapid adaptation process for directions, numbers, and letters: Using TS-VFC learning to form LTS, with pre-training conducted through random wrist movements, followed by fine-tuning with few-shot labeled data for quick application across various tasks.