Development of an AI Ecosystem for AR Applications Within Aerospace

The purpose of this paper is to develop an Artificial Intelligence (AI) ecosystem, to effectively introduce a dynamic approach for Augmented Reality (AR) based instruction modules through predictive machine learning algorithms. The main applications explored are the manufacturing and maintenance sectors within the Aerospace industry. This paper explores the possibility of implementing an innovative process of integrating efficient data handling methodologies and machine learning techniques through a micro-level coordinate system approach. ML algorithms which are expected to have high prediction accuracies have been analyzed and a proof of concept prototype has been developed with a selected algorithm. There are various applications within AR based manufacturing/maintenance instructions in the Aerospace industry, the use cases span from aircraft part manufacturing & maintenance, CAD modelling, part assembly and routine maintenance within the International Space Station (ISS). The research study in this thesis looks to optimize the current AR model-based animation techniques by incorporating AI methodologies throughout the overall process. The current techniques include: Using 3D image information from calibration markers, Teleconference setting for improved communication and Mobile viewing & prototyping unknown surfaces using the HoloLens. This paper proposes an innovative methodology that current techniques can adapt to improve efficiencies and capabilities