A Biomimetic Binocular Visual System Based on β ‐Ga 2 O 3 Optoelectronic Synaptic Device for Motion Recognition

Abstract

ABSTRACT With the rapid development of autonomous vehicles, robotics, and monitoring systems, the demand for having visual system capable of recognizing explicit motion information, such as motion direction, state, and speed, has increased substantially. To address this demand, biomimetic visual systems have garnered considerable attention owing to their high speed and capability of collecting intricate motion details. However, most biomimetic visual systems are limited to two‐dimensional (2D) recognition via monocular manner. In contrast, binocular visual systems could extend the recognition scope from the 2D plane to three‐dimensional (3D) space. This study proposes a biomimetic binocular visual system for 3D motion recognition based on β ‐Ga 2 O 3 synaptic devices to gather motion information including motion direction, state, and speed. Based on trap behaviors, the broad tunability of synaptic behavior endows β ‐Ga 2 O 3 synaptic device with multiple functions including photodetector (PD), compressor and activation function generator (AFG), realizing the recognition of nineteen directions in 3D space with a high accuracy of 98.83%. Furthermore, two additional applications, namely motion state and speed recognition, are also validated by this system. These results pioneer the development of a biomimetic binocular visual system and paves a solid path for future 3D motion recognition.