Stretchable and stable neuromorphic tactile system

Abstract

Artificial neuromorphic tactile systems based on tactile sensors and artificial neurons can simulate biological perception systems and achieve the efficient transmission, processing and utilization of information. However, limited by the materials and structures of devices, it is difficult for artificial neuromorphic tactile systems to have stable perception abilities such as biological functions, especially when the devices are bending or stretching. In this work, we propose a performance-stable artificial tactile neuron, which integrates a stretch-insensitive triboelectric nanogenerator (TENG) with an artificial neuron in a single device. The stable contact triboelectric output could stimulate a neuron and affect its firing time. Based on this principle, we established a 64 × 64 neuromorphic tactile matrix, which processed the triboelectric signal generated by touch through the integrate-and-fire neuron and simulated the pressure trajectory and texture of tactile perception based on the firing time classification. This provides an effective strategy for the future simulation of biological skin perception and fast classification and recognition of information.