Multilayer Bionic Tunable Strain Sensor with Mutually Non‐Interfering Conductive Networks for Machine Learning‐Assisted Gesture Recognition
Liang Gao, Jie Yang, Yi Zhao, Xinxin Zhao, Kangkang Zhou, Wei Zhai, Guoqiang Zheng, Kun Dai, Chuntai Liu, Changyu Shen
Abstract
Abstract Flexible strain sensors capable of acquiring tiny mechanical signals and attaching to various irregular surfaces are becoming prevalent in physiological measurement, soft robotics, and human‐machine interaction. However, the advancement of flexible strain sensors is substantially impeded by the intrinsic compromise between sensitivity and the range of detectable signals. Inspired by the multilayer structure of nacre in nature, a carbon nanotubes (CNTs)/graphene (GR)/graphene (GR)/thermoplastic polyurethane (TPU) mat (CGGTM) is prepared by electrospinning and high‐pressure spraying technology. By designing a multilayer structure with mutually non‐interfering conductive networks, the resulting CGGTM possesses a low detection limit (0.05% strain), high sensitivity (gauge factor, GF > 152537), large detection range (up to 364% strain), fast response/recovery time (80 ms/100 ms), and excellent cyclic durability (up to 1000 cycles). Furthermore, the CGGTM also exhibits satisfactory triboelectric performances when assembled into a triboelectric nanogenerator (TENG, 3 × 3 cm 2 ), including high triboelectric output (open‐circuit voltage V oc = 135.4 V, short‐circuit current I sc = 1.25 µA) and power density (88 mW m −2 ), enabling reliable power supply, self‐powered sensing, and pulse monitoring capability. Finally, CGGTM is successfully applied to the collection of biological signals and multi‐gesture motion recognition assisted by machine learning algorithms, which holds promise for intelligent interaction with gestures in the future.