Annotating 3D Scenes with Knowledge Graphs for Smart Infrastructure Digital Twins

The accurate semantic representation of 3D assets is increasingly crucial for the development of Digital Twins (DTs) in smart infrastructure scenarios. Although standard formats like Universal Scene Description (USD) provide a robust framework for modeling and composing large-scale 3D environments, they lack native support for expressing ontology-based semantics. This paper introduces a lightweight ontology-agnostic framework to embed annotations based on Web Ontology Language (OWL) directly within USD scenes. The proposed approach enables the association of OWL individuals, class expressions, and property assertions with scene primitives, supporting reasoning and integration with knowledge graphs without altering geometric or rendering behavior. The framework is designed to be modular and general-purpose, enabling deployment across authoring pipelines and visualization platforms. A prototypical implementation in the Unity engine demonstrates the feasibility of the approach and highlights its potential to enrich smart infrastructure DT representations with machine-understandable semantics.