Dynamic keystroke-password recognition based on piezoelectric-triboelectric coupling sensor array with crosstalk-free for authentication system

As artificial intelligence technologies progress, Human-Machine Interactions (HMI) must evolve rapidly, necessitating reliable and continuous authentication solutions. We propose dynamic keystroke pattern recognition technology based on a piezoelectric-triboelectric coupling sensor array to address these challenges, enhancing the keystroke signal characteristics. Dual verification technology combining password and biometric authentication effectively enhances the security level of the Human-Machine Interactions system. The triboelectric sensor efficiently reduces the output channels of a 3 × 3 sensing array to a single channel using a mesh topology electrode design. Each of the nine triboelectric sensor units corresponds to nine numeric key, allowing users to input different key combinations that generate unique cryptographic waveforms distinguished by individual keystroke characteristics. A piezoelectric-triboelectric coupling sensor array, structured with multiple layers, is devised. By incorporating a piezoelectric sensor in the upper layer, we harness the complementary effects of piezoelectric and triboelectric properties to boost authentication accuracy and alleviate the limitations of single sensing modalities. Notably, crosstalk is eliminated through the specialized sensor array and the topological electrode design. Integrating the piezoelectric-triboelectric coupling sensor array with a 1D CNN neural network approach achieves password recognition accuracy surpassing 99%, effectively mitigating the risk of password leakage in systems facilitating human-computer interactions.