Self-Powered Handwritten\nLetter Recognition Based\non a Masked Triboelectric Nanogenerator for Intelligent Personal Protective\nEquipment

As one of the most important ways of human–machine\ninterfaces,\nthe touchpad has excellent input convenience. Input devices for extreme\nenvironments require simpler structures and diverse inputs to ensure\ninformation inputs. This paper proposed a self-powered flexible input\npanel with single-channel output for the input recognition of all\n26 letters, and a paper mask was implemented to cover the triboelectric\nnanogenerator (TENG) board and obtain more complicated electrical\nsignal features. Based on the change of the triboelectric output of\nthe mask, neural network models with different combinations of layers\nwere designed and optimized, and the highest recognition rate of 88.7%\nfor all letters and 100% recognition accuracy for some letters were\nachieved among the five testers. For letters with low recognition\nrates, a specific writing specification was further proposed to improve\nthe accuracy of model recognition. These results facilitate the application\nof the proposed input panel as a flexible wearable device and personal\nprotective equipment for extreme environments including chemical,\nbiological, radiological, nuclear (CBRN) scenarios or aerospace.