Self-powered perception system based on triboelectric nanogenerator and artificial neuron for fast-speed multilevel feature recognition

Abstract

Inspired by information processing in biological systems, the data processing efficiency of sensor systems combined with artificial neuromorphic devices has been significantly improved. However, the current reports on artificial perception systems are mainly based on artificial synapses, while little attention has been taken on the artificial neuron-based perception systems. Here, we propose a self-powered sensory neuron (SPSN) composed of a TaOx-based device-level artificial neuron and a high-sensitivity triboelectric nanogenerator (TENG) for self-powered artificial tactile sensing system. The SPSN with volatile switching and a high on/off ratio of 105 can mimic the leaky-integrate-and-fire (LIF) neuron model without additional circuitry and reset operations, while multiple stimulus inputs can be integrated to extract characteristic events due to its corresponding firing times to different stimuli. Moreover, we successfully constructed a 64´64 neuron-based artificial sensing array system and extracted the pressing trajectories and textures by the dynamic threshold characteristics of sensory neurons. Compared with the synapse-based artificial perception system, the neuron-based artificial perception system provides more feature extraction layers and faster data processing speed. Our results provide an effective strategy for building next-generation neuromorphic perception networks, intelligent human-computer interaction and high-speed feature recognition.