High-Performance Transparent, Deformable, and Recoverable Biomimetic Stevia-PVA Hydrogel Triboelectric Nanogenerator with Machine Learning-Assisted Motion Recognition.

Rapid breakthroughs in IoT and AI have raised demand for portable, self-powered, flexible sensor devices. Hydrogels with high conductivity, mechanical tunability, environmental adaptability, and biocompatibility are a clever way to develop flexible sensors for triboelectric nanogenerators. TENG application is limited by the paucity of suitable biomaterials and the need for highly conductive fillers such 2D materials, which trade off transparency, output, and sensing. A unique, very transparent, highly stretchable, high-output performance biomimetic stevia/PVA hydrogel-based triboelectric nanogenerator (S-TENG) is investigated to overcome this issue. Due to its abundant dynamic hydrogen bonding, cost-effective biomimetic stevia is added to polyvinyl alcohol (PVA) to increase hydrogel cross-linking and crystalline domains. These structural advancements give the S-hydrogel 2-5 times the mechanical strength and 3-8 times the electrical output of 2D-, bio-, and transparent-material-based TENGs, while maintaining transparency. The S-hydrogel may be recycled and recovered by water-assisted dissolution and re-gelation, keeping its voltage output. The improved S-TENG is a self-powered sensor for various human motions with great sensitivity and a 13-ms reaction time. The XGBoost method had the greatest classification accuracy of 95.29% among eleven machine learning models, showing the promise of self-powered sensors for many applications.