Ultrasensitive Flexible\nStrain Sensor Made with Carboxymethyl-Cellulose-Anchored\nCarbon Nanotubes/MXene for Machine-Learning-Assisted Handwriting Recognition

Abstract

The combination of wearable sensors with machine learning\nenables\nintelligent perception in human–machine interaction and healthcare,\nbut achieving high sensitivity and a wide working range in flexible\nstrain sensors for signal acquisition and accurate recognition remains\nchallenging. Herein, we introduced carboxymethyl cellulose (CMC) into\na carbon nanotubes (CNTs)/MXene hybrid network, forming tight anchoring\namong the conductive materials and, thus, bringing enhanced interaction.\nThe silicone-rubber-encapsulated CMC-anchored CNTs/MXene (CCM) strain\nsensor exhibits an excellent sensitivity (maximum gauge factor up\nto 71 294), wide working range (200%), ultralow detection\nlimit (0.05%), and outstanding durability (over 10 000 cycles),\nwhich is superior to most of the recently reported counterparts also\nbased on a conductive composite film. Moreover, the sensor achieves\nseamless integration with human skin with the help of a poly(acrylic\nacid) adhesive layer, successfully obtaining stable and clear waveforms\nwith meaningful profiles from the human body. On this basis, we proposed\nand realized a novel in-air handwriting recognition method via extracting\nmultiple features of high-quality strain signals assisted by deep\nneural networks, achieving a high classification accuracy of 98.00\nand 94.85% for Arabic numerals and letters, respectively. Our work\nprovides an effective approach for significantly improving strain\nsensing performance, thereby facilitating innovative applications\nof flexible sensors.