Beyond the pixel: Artificial vision based on optoelectronic memristor arrays

Advanced optoelectronic integration with sensing, storage, and computing functions shows great potential in next generation of vision hardware. To approach pathbreaking hardware design for in-situ multifunction and high parallelism, the rise of optoelectronic memristor offers a highly promising avenue for accelerating functional development of vision hardware. This paper provided new perspective of basic optoelectronic memristors characteristic and integration, array function in various visual level and their application roadmap in artificial vision systems. Firstly, the working principle and modulation classification of optoelectronic memristors has been detailly reviewed, and we further explored device physical models as well as essential artificial synaptic characteristics. Secondly, we introduced the physical architectures of optoelectronic memristor arrays and ANN computational frameworks for constructing artificial visual systems. Thirdly, the applications of integrated optoelectronic memristor arrays in artificial visual systems and their vision function from perception and memory, low-level preprocessing and high-level computing were discussed. Finally, we discussed the future prospects and current challenges of the integrated optoelectronic memristor arrays, which represent a key focus for future research and require collaborative efforts from researches.