Digital twin scene descriptor : a standardized schema for dynamic IoT-driven digital-twin generation and operation
Loading...
Date
2025-06-30
Authors
Στεργιόπουλος, Γεώργιος
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Digital Twins (DTs) have become a key tool for monitoring, analyzing, and optimizing complex systems, yet most platforms handle 3D geometry, IoT telemetry, and simulation data in ad-hoc, often proprietary, workflows that hinder interoperability and real-time interaction. This thesis proposes the Digital Twin Scene Descriptor that treats 3D assets, context entities and update channels as first-class, loosely coupled resources and builds a prototype implementation of it based on a JSON/NGSI-v2 standard. Building on DTSD, we implement (i) a plug-in for Three.js web viewer able to translate the description entities into an interactive, browser-based twin that streams live spatial and contextual updates and representation assets, (ii) a SUMO integration layer that ingests real-car positions and projects virtual traffic back into the twin, thus closing the cyber-physical loop and (iii) a Model Context Protocol based transformative automation pipeline in which a large-language model autogenerates the DTSD entities and their needed resources, including 3D assets, for existing IoT entities. Demonstrations show real-time projection of GPS-tracked physical objects into the DT, live collision avoidance between real and simulated cars and one-click integration of an NGSI parking sensor into the twin complete with geometry synthesis and data binding. The results confirm that DTSD enables real-time bidirectional synchronization, seamless simulator federation and automated twin creation, laying the groundwork for large-scale, vendor-agnostic Digital Twins.
Description
Keywords
Digital twins, Digital twin scene descriptor, NGSI, FIWARE, Three.js, 3D Web viewer, SUMO traffic simulator, Model context protocol, Large language models, Gaussian splats