In-sensor Computing Based on Two-terminal Optoelectronic Memristors

Abstract

In-sensor computing, which is capable of combining sensing, processing, and memory in a single unit, has been proposed to implement a brain-inspired artificial visual system. Among the proposed candidates, two-terminal optoelectronic memristors with optical and electrical hybrid stimuli have been considered to be one suitable hardware for in-sensor computing due to their low power consumption, fast operating speed, and high scalability. Up to now, various optoelectronic memristive devices have been developed to overcome the drawbacks of conventional artificial visual systems with partition of sensory, memory and processing units. In this chapter, the recent progress in two-terminal optoelectronic memristors for in-sensor computing is focused. We introduced the physical mechanism of optoelectronic memristors in detail, including photon–electron and photon–ion coupling processes. The main operation behaviors can be classified into light–electric hybrid-controlled and fully-light controlled types. Moreover, optoelectronic applications, such as optoelectronic logic, image detection and memorizing, and low/high-level sensory processing functions, were also presented. Finally, a brief discussion of the perspectives and challenges for further development of in-sensor computing based on optoelectronic memristors is presented.