In-sensor Computing Devices for Bio-inspired Vision Sensors

Abstract

The rapid development of Internet of Things generates abundant data, which demands a new computing paradigm to efficiently process these data at sensory terminals. Vision is the most typical sensor-data-intensive application. Biological visual systems have considerable advantages in terms of energy efficiency and multiple functionalities. The existing artificial vision system based on conventional image sensors, memory, and processing units has complex circuitry, high power consumption, and limited perception range. We design and demonstrate optoelectronic devices for bio-inspired visual sensors, which enable image sensing, neuromorphic visual pre-processing, and effective adaptation to a wide range of light-intensity.