Deep-Learning Enabled Active Biomimetic Multifunctional Hydrogel Electronic Skin

Abstract

There is huge demand for recreating human skin with the functions of epidermis and dermis for interactions with the physical world. Herein, a biomimetic, ultrasensitive, and multifunctional hydrogel-based electronic skin (BHES) was proposed. Its epidermis function was mimicked using poly(ethylene terephthalate) with nanoscale wrinkles, enabling accurate identification of materials through the capabilities to gain/lose electrons during contact electrification. Internal mechanoreceptor was mimicked by interdigital silver electrodes with stick-slip sensing capabilities to identify textures/roughness. The dermis function was mimicked by patterned microcone hydrogel, achieving pressure sensors with high sensitivity (17.32 mV/Pa), large pressure range (20-5000 Pa), low detection limit, and fast response (10 ms)/recovery time (17 ms). Assisted by deep learning, this BHES achieved high accuracy and minimized interference in identifying materials (95.00% for 10 materials) and textures (97.20% for four roughness cases). By integrating signal acquisition/processing circuits, a wearable drone control system was demonstrated with three-degree-of-freedom movement and enormous potentials for soft robots, self-powered human-machine interaction interfaces of digital twins.