Handwriting to Digital Text Translation Using a Self-Powered Triboelectricity-Induced Piezoelectric Writing Pad through a Deep Learning Algorithm

Human–machine interfaces require energy generation, communication, and computing power to complete the assigned tasks. Here, we introduce a triboelectricity-induced piezoelectric writing pad (TPWP) for advanced letter detection technology capable of converting handwritten inputs into digital text. The device not only preserves the tactile and intuitive nature of traditional writing but also harnesses the efficiency of digital text, offering users the advantage of instant and accurate transcription for enhanced productivity and streamlined information management. The efficacy for handwriting conversion is demonstrated by the acquisition of handwritten signals from three individuals encompassing English letters “ A ”, “ B ”, “ C ”, and “ D ” employing the micropattern TPWP. These signals exhibit detectable unique features for the same letter written by different people and recognition of a similar pattern for a particular letter. To enhance the accuracy of letter recognition, a handwritten pattern was filtered and trained up to 250 epochs. The time-varying signal was transformed into a spectrogram using a deep learning algorithm. Convolutional neural network (CNN) integration allows the deep learning algorithm to successfully detect each letter. High classification accuracy is reached with letters “ A ” and “ B ” scoring 100%, while “ C ” and “ D ” score 99%. Overall, the suggested model’s precision is 99%. Our results reveal the potential of the 3D printed textured TPWP for handwriting to text conversion applications.