Nexus2030 takes shape: models, data and artificial intelligence
During the project’s second year, efforts focused on integrating developed components and implementing the first technological prototypes.
Within WP4, predictive models based on System Dynamics were transformed into operational simulation tools, integrated into an interactive runtime environment, accessible even to users with basic skills. This enabled the creation of forecasting scenarios to support territorial planning.
WP5 consolidated the Digital Twin architecture, defining each functional layer (data, ontologies, APIs, AI modules, visualization), and completed the implementation of the data lake and common data space, where information can flow and be aggregated according to semantic criteria.
Meanwhile, WP3 led to the development of user-friendly interfaces based on Jupyter Notebooks, enabling the creation, training, and explanation of AI models through the integration of AutoML frameworks (AutoGluon, Auto-sklearn) and XAI modules that ensure output interpretability.
Thanks to WP6, the first prototypes of interactive urban data visualization were developed, including 3D dashboards based on GIS and BIM tools for building exploration. Simultaneously, WP7 introduced the first user interaction tools (Information Space and Simulation Space), with features such as personalized scenario generation, smart notifications, and augmented reality visualizations.
Highlights:
- Runtime environment for customizable territorial simulations
- Complete Digital Twin architecture and operational common data space
- AutoML and XAI libraries integrated with intuitive Jupyter interfaces
- GIS/BIM visualizations on dashboards and AR interaction prototypes