Unit-Area Optimized Flexible Triboelectric Nanogenerators for High-Performance Human Motion Sensing With Deep Learning

Triboelectric nanogenerators (TENGs) have garnered significant attention as self-powered energy harvesters and sensors, yet their performance critically depends on structural design and material engineering. In this work, we introduce a structured TENG designed with the concept of unit-area to maximize contact efficiency by minimizing mechanical interference between neighboring structures. Synergistically, barium titanate (BaTiO₃) nanoparticles were incorporated into the elastomer matrix to enhance the dielectric constant and surface charge density. Based on this approach, unit-area-based TENGs (UA-TENGs) with various geometries were comprehensively evaluated through computational analysis and experimental validation. The results reveal that smaller structures enable more effective contact, while dome structures exhibit superior output compared to pyramid structures. Quantitatively, the optimized dome-structured UA-TENG with the BaTiO₃ composite achieves a power density of 18 W/m², a maximum voltage of 254 V, and enhanced pressure sensitivities. The device successfully powered 127 LEDs and operated small electronics. Furthermore, the UA-TENG demonstrated excellent performance in detecting biomechanical signals. Integration into a smart glove enabled real-time gesture recognition, where a convolutional neural network accurately classified five distinct hand motions to achieve reliable wireless drone control. These results demonstrate the broad application prospects of the UA-TENG in wearable energy supply, intelligent sensing, and human-machine interaction.