講演情報
[C-12-43]Development of a Preprocessing-Free Light-to-Digital Trans-Nail PPG Sensor for Wearable Controllers
◎△Pattaramon Thianmontri1, Aoi Kataura2, Yuqi Duan1, Akihiro Suzuki2, Ryotaro Kawashima2, Kotaro Torazawa1, Yukino Sakurai1, Takafumi Fukushima1,2, Koji Kiyoyama3, Tetsu Tanaka1,2 (1. Graduate School of Engineering, Tohoku University, 2. Graduate School of Biomedical Engineering, Tohoku University, 3. Graduate School of Engineering, Nagasaki Institute of Applied Science)
キーワード：
Photoplethysmography、Reservoir Computing、Controller
This work presents a preprocessing-free, light-to-digital trans-nail photopletysmography (PPG) sensor framework for edge wearable controllers. Utilizing transmitted-mode PPG across the nail bed, the system tracks pressure-induced changes in subungual blood volume against the rigid nail plate to capture user intent. The proposed custom sensor chip integrates an on-chip photodiode and an ambient light cancellation (ALC) circuit to eliminate noise. The resulting signal directly modulates a light-to-frequency converter, translating force-induced variations into proportional pulse trains. These shifts are captured directly as 16-bit counter values, completely eliminating software-side filtering or feature extraction. For validation, human-subject experiments were conducted to classify four states: idle (no-contact), touch, press, and hard press. Feeding the raw 16-bit digital stream directly into a lightweight reservoir computing (RC) network, the system achieved an overall classification accuracy of 90.28%. This streamlined sensor-processor co-design minimizes computational overhead, providing a highly responsive, low-power interface ideal for real-time edge applications.