A Non-Contact Triboelectric Sensing System and Its Application in Material Identification

In the interaction between artificial devices and the world, sensing serves as a bridge that grants artificial objects the ability to perceive and understand the world. Here, we propose a non-contact triboelectric sensing system that combines triboelectric nanogenerators with machine learning. This system is characterized by its ability to obtain quantitative parameters of the measured object without direct contact and ensures long-term stability. The sensing system integrates a triple triboelectric sensing array, and by combining triboelectric sensing with machine learning, it achieves precise identification of material types in an open environment. Theoretically, since each material has a different ability to gain or lose electrons, a special triboelectric waveform signal is generated due to electrostatic induction when the triboelectric sensing unit approaches the measurement object. By using a three-channel sensing module to reduce environmental interference and integrating a convolutional neural network model, the system achieves a $\mathbf{9 6. 2 5 \%}$ accuracy rate in material identification tasks. This sensing system provides significant reference value for non-destructive, clean detection in the fields of medical health and intelligent manufacturing.