Desenvolvimento de algoritmo para alinhamento de fibra ótica com guia de onda em dispositivo planar

Abstract

This work has the goal of developing an algorithm, using image processing and machine learning techniques, to control an alignment plataform in order to obtain a low coupling loss between optical fibers and waveguides of photonics circuits, with resolution in the order of nanometers, replacing the alignment made by human technicians. The alignment algorithm is split in two parts, an initial alignment (rough estimate) based on images, followed by a more precise alignment, using power meters. The focus of the project is in the initial alignment, which is essential to reduce the search space for the next step. Without this reduction the precise alignment may not finish. This step of alignment is an important challenge mainly because of the reduced size of the components and the precision needed, both of which making it impossible for simple methods of image processing to work properly. The algorithm of precise alignment has been created elsewhere and will be used with just small changes to operate within this work.