Electrode Array-Free Tactile Sensor for Addressable Force Sensing Assisted by a Neural Network

Abstract

Recently, tactile sensors have attracted great attention for artificial skin, medical surgery, and robotic manipulation applications. To implement large-area force value and location sensing, vast electric wires and arrays are required to be embedded in sensors, leading to difficulties in system integration, electrical signal transport, and analysis. Herein, a neural network-assisted electrode array-free capacitive sensor is proposed, which only requires four wires for sensing areas of any scale. Four eutectic gallium-indium (EGaIn)/poly(dimethylsiloxane) (PDMS) composite films are used as the electrodes allowing high flexibility and electrical conductivity. The dielectric layer of the capacitor is designed with gradient thickness to implement the one-dimensional stimulus's location detection so the two-dimensional coordinate can be detected by integrating two sensing units. With the assistance of a neural network, the value and location of force signals can then be analyzed and predicted according to the capacitance variation of two sensing units, which has been verified by the experimental results. The developed sensing device exhibits a simple and low-cost model for tactile sensors, providing insights into tactile sensor design.