人工神经网络
反向传播
标度系数
应变计
计算机科学
方向(向量空间)
材料科学
强度(物理)
干扰(通信)
声学
图层(电子)
机器人
电子工程
人工智能
电子皮肤
感觉系统
拉伤
生物系统
横截面
机器人学
信号(编程语言)
传感器阵列
光电子学
生物医学工程
作者
Xuewei Shi,Alamusi Lee,Yaolu Liu,Chenxing Xiang,Huiming Ning,Haowen Liu,Hansheng Liao,Kexu An,Bo Yang,Kaiyan Huang,Bin Gu,Ning Hu
标识
DOI:10.1002/admt.202502049
摘要
ABSTRACT Mechanically stretchable strain‐sensing electronic skin (E‐skin) has garnered significant attention for bioinspired sensory systems and human‐machine interfaces (HMI). However, conventional E‐skin devices or arrays remain limited in achieving high‐precision simultaneous detection of strain direction and intensity, which holds promise for enabling robots to recognize touch gestures, and even interpret intentions and emotions. To address this challenge, this study developed a heterogeneous E‐skin that integrates three stretchable strain‐sensing units with a backpropagation (BP) neural network algorithm. Each sensing unit features a heterogeneous fence‐structured design, which minimizes transverse strain interference and endows orientation sensitivity‐responding exclusively to longitudinal strains. The sensory layer exhibits excellent flexibility, stretchability, and cyclic stability, with a gauge factor (GF) of 2.96 under 0–50% strain, and 10.93 under 50–80% strain. Through combined simulation analysis and experimental validation, the heterogeneous structure was optimized to achieve superior directional strain response characteristics. Notably, by employing the BP neural network to decode correlated signals from the three sensing units, the E‐skin achieves high‐precision detection of both the direction and intensity of an arbitrary applied strain.
科研通智能强力驱动
Strongly Powered by AbleSci AI