Halide Perovskite Retinomorphic In-Sensor Computing

The processing efficiency of the retinal sensory system can be attributed to its internal microcircuits. They are capable of encoding color information, enhancing edge information, adapting to dynamic lighting, detecting motion as well as learning from experiences. Low-power, compact design configurations make perovskite-based sensor devices highly attractive for in-sensor computing applications. We discuss how the richness of structural tuning and operational sensitivity in halide perovskites present a promising opportunity to emulate retina’s sensing, processing and information storage capabilities. Contrary to conventional broadband semiconductors, semitransparent halide perovskites offer interesting in-sensor computing device configurations such as vertically stacked photodetectors and photocapacitive systems. From memdiodes, filament switches to charge-trapping transistors, we discuss the realization of low-powered memory and the significance of self-powered photodetectors. Lastly, we examine retinal features which can be realized by various retinomorphic implementations and discuss how we can build a memory-imbued retina-inspired vision system.