Arch-Bridge-Inspired Auxetic Piezoelectric Nanogenerators for Machine Learning-Assisted Object Recognition

Piezoelectric polymers are increasingly in demand for wearable devices and soft robotics, where high sensitivity, tunability, and versatile mechanical responses are essential. However, the application of piezoelectric sensors is constrained by inadequate piezoelectric output and deformation-induced unstable signal collection. Inspired by arch bridges as well as guided by finite element analysis, the study successfully designed and fabricated a piezoelectric metamaterial with a negative Poisson's ratio. Due to the unique architecture, the piezoelectric response in bending mode is significantly enhanced. Simultaneously, the piezoelectric coefficient increases from 11.7 to 23.8 mV/V through the incorporation of modified montmorillonite as a functional dopant, combined with advanced Fused Deposition Modeling techniques, significantly improving the piezoelectric response under impact mode. The novel I-shaped piezoelectric nanogenerator not only achieves efficient energy harvesting but also functions as a high-sensitivity flexible sensor capable of accurate object recognition. When integrated with a Convolutional Neural Network (CNN), the assembled system could achieve a classification accuracy of up to 98.1%. This work not only expands the application potential of piezoelectric polymers in multifunctional flexible electronics but also provides a novel design paradigm for next-generation self-powered intelligent systems.