Fish-Scale Sensors Power a Glove That Translates Sign Language
Researchers made a flexible glove by treating discarded fish scales, heating them, and embedding the resulting carbon-rich powder in a polymer. The glove’s nine sensors generate voltage as the hand moves, and a neural network translates the patterns into six sign-language words with about 94% accuracy. The self-powered design avoids bulky batteries and may offer privacy advantages over camera-based systems.
Researchers converted discarded fish scales into carbon-rich powder by acid treatment and heating to 450°C, then dispersed it in PVDF nanofibers within stretchy silicone. At 0.14% by weight, output improved and crystallization changed; excess powder clumped and leaked charge. The film stretched about 49% and survived roughly 10,000 cycles, sensing 0.25–12.5 kPa.
The glove places nine sensors on joints, palm, hand back, and wrist. Its triboelectric effect turns motion into voltage, which a spatial-temporal graph convolutional network classifies as six signs or idle, reaching 94.43% accuracy. Earlier gesture gloves used yarn sensors or triboelectric signals, while camera systems avoid wearables but need clear views and raise privacy issues.
This prototype could eventually benefit deaf and hard-of-hearing signers by offering a wearable, camera-free route to basic machine translation, potentially in low-light or privacy-sensitive settings. Its self-powered sensing may also reduce battery bulk, making everyday use more practical. However, with only six words, idle detection, and prototype-level testing, it may remain a research step rather than a substitute for interpreters or fluent conversation. Developers and privacy-conscious users may watch the approach, but real impact depends on vocabulary, robustness, and user acceptance.