A Polychromatic Neuromorphic Visual System Inspired by Biomimetics for Miniature Insect Robots

Abstract

The emergence of electronics influenced by visual neural perception and action is increasingly crucial for enhancing interactive human-machine interfaces and advancing the capabilities of intelligent robots. There is an urgent demand for a system that incorporates neuromorphic environmental information encoding, synaptic signal processing, and motion control. Taking inspiration from the polychromatic visual system, it is initially employed bulk heterojunction organic photosynapses (BHJ-OPS) to replicate the functionalities of human-like visual nerve system. The BHJ-OPS, utilizing a two-terminal architecture, exhibits an ultra-broadband photodetection range (365-1060 nm). For near-infrared (NIR) perception, an optical energy consumption as low as 0.2 fJ per synaptic event is demonstrated, which is the lowest energy consumption achieved so far with NIR light stimulation. By combining the photovoltaic effect in heterojunctions with electron trapping in the buffer layer, BHJ-OPS displays bio-synaptic characteristics such as short-term and long-term memory, as well as experiential learning, which endows the synapse array with multispectral color-discrimination capabilities. Finally, it is implemented miniature insect robots capable of night-time foraging and predator evasion based on a simulated 26 × 26 memristor network. This demonstrates significant potential for the development of miniature insect robots with self-regulation and adaptability, particularly in exploration, monitoring, and rescue missions.