Optical neural networks for low-latency and energy efficient applications in production

Novel photonic compute hardware is a rising and promising technology wherever low-latency or energy efficient computation is required.Especially, optical neural networks (ONNs) aim to provide accelerators for artificial intelligence (AI) applications.In this work, we present a prototype of an optical matrix-vector multiplier.As the technology is still in its infancy, we motivate an outlook on future performances.Furthermore, we map our technology to the field of (automotive) production, where ONNs may be applied in the future.Hereby, we take into account the Profinet communication protocol, which is widely used by German car manufactures.This paper manifests a proposal for future applications of ONNs in production and its logistics.