Towards a data fabric framework for industrial metaverse integration

Abstract

The integration of digital and physical assets in industrial settings is increasingly facilitated by the concept of the Industrial Metaverse, a unified platform that leverages advanced technologies like AR/VR/XR to create interconnected 3D environments. As industries within supply chains become more interlinked, the need for seamless integration across these digital environments becomes critical. This paper addresses the challenge of interconnecting private industrial metaverses by proposing a novel data fabric framework that supports diverse modalities, ensures privacy, and adheres to business agreements. We highlight the role of the distributed compute continuum in the framework and demonstrate its practical utility through deployments in Unity and Omniverse across various geographic locations, highlighting the need for AI-based interconnection to optimize real-time analytics and operational scalability. Our experiment shows that even transmitting a 3D patch file, which is considerably smaller than the original base file, across metaverses in different geographic locations requires substantial computational and communication resources, potentially limiting real-time collaboration between metaverses. This observation highlights the importance of AI-based interconnection in enabling a textual metaverse, where instead of transmitting 3D patch objects over the web, only a fine-grained textual description of the patch file is shared. This method allows for more efficient transmission using current networking technology. The textual metaverse is expected to employ an AI-based encoder at the source and an AI-based decoder at the destination to convert the 3D patch into text and subsequently reconstruct it back into the 3D patch.