Oscar Roman

Recent advances in 3D reconstruction have enabled the automated generation of city models from LiDAR point clouds and aerial imagery. Although these methods produce visually convincing models, they are primarily designed for visualization purposes and often lack the geometric and topological quality required for Urban Digital Twin applications, such as physics-based simulation, semantic analysis and predictive modelling. Automatically reconstructed LoD2 building models frequently exhibit defects including open boundaries, non-manifold configurations, inconsistent face orientations and degenerate geometries. This work presents a graph-guided framework for the topological repair of LoD2 building models. Mesh connectivity is represented as a graph, allowing boundary, adjacency and local geometric information to identify defective regions and guide robust repair operations. By combining topology-aware reasoning with geometry-constrained graph processing, the proposed method restores watertightness, improves 2-manifold consistency and preserves the structural integrity of complex building surfaces. Unlike conventional geometry-based post-processing techniques, the proposed approach explicitly targets topological correctness, producing high-quality building models suitable for simulation and analysis within Urban Digital Twins.

Digital Twins in the Architecture, Engineering, and Construction (AEC) domain support monitoring, simulation, and increasing levels of automation in building management across scales. Energy Digital Twins are particularly demanding, requiring (i) simulation-grade geometry and (ii) persistent topology and semantics across monitoring- and scenario-driven updates. This paper proposes a unified multi-representation EDT in which (i) a watertight, solid, and (ii) a topology-preserving B-Rep are co-maintained through a mapping layer that preserves object identity and links geometry to a typed property graph. Building on this, the presented Scan-to-Energy Digital Twin pipeline converts raw point clouds into multi-level EDT instances by integrating Scan-to-BIM reconstruction, topological modelling, semantic enrichment and parser–transformer–writer interoperability modules. The graph-backed EDT enables reversible export to epJSON and gbXML (optionally IFC), supporting scenario-based EnergyPlus simulations and incremental retrofit updates, such as insulation thickness and window thermal transmittance value changes. Validation on a set of four buildings achieves 0.86–0.89 mAPv and schema-valid exports, demonstrating the effectiveness of our end-to-end approach for interoperable energy analysis, monitoring, and operational decision support.

In response to the increasing demand for effective bridge management and the shortcomings of current proprietary solutions, this work presents an open-source, web-based platform designed to support bridge inspection and data management, particularly for small and medium-sized public administrations, which often lack personnel or funding for implementing context-specific tools. The system addresses fragmented workflows by integrating multi-format geospatial and 3D data—such as point clouds, CAD/BIM models, and georeferenced imagery—within a unified, modular architecture. The platform enables structured inventory, interactive 2D/3D visualization, defect annotation, and role-based user interaction, aligning with FAIR principles and interoperability standards. Built entirely with free and open-source tools, the P.O.N.T.I. prototype ensures scalability, transparency, and adaptability. A multi-layer navigation interface guides users through asset exploration, inspection history, and immersive 3D viewers. Fully documented and publicly available on GitHub, the system allows for deployment across varying institutional contexts. The platform’s design anticipates future developments, including integration with IoT monitoring systems, AI-driven inspection tools, and chatbot interfaces for natural language querying. By overcoming existing proprietary limitations and providing access to a versatile single space, the proposed solution supports decision-makers in the digital transition towards a more accessible, transparent and integrated infrastructure asset management.

Digital Twins (DTs) are transforming construction and energy management sectors by integrating 3D surveying, monitoring, Building Performance Simulation (BPS), and Building Energy Simulation (BES) from the earliest design or retrofit stages. Moreover, dynamic thermal simulations further support energy performance assessments by modeling indoor conditions to meet comfort and efficiency targets. However, their reliability depends on accurate, standards-compliant 3D building models, which are costly to create. This research introduces a complete framework for automatically generating energy-focused Digital Twins (EDTs) directly from unstructured point clouds. Combining Deep Learning-based instance detection, Scan-to-BIM techniques, and computational geometry, the method produces simulation-ready models without manual intervention. The resulting EDTs streamline early-stage performance evaluation, enable scenario testing, and enhance decision making for energy-efficient retrofits, advancing smart-building design through predictive simulation.

[ENG] The research addresses the theme of the digital representation of existing buildings and prefabricated components, exploring the relationships between survey restitution, analysis of the current state, design of interventions, and production. The focus has centered on the different possible uses of digital models to simplify workflows, by correlating two previous experiments: the development of parametric families for the representation of prefabricated panels used to envelope and improve the performance of existing buildings; the study of modelling strategies for the representation of the current state of an existing residential building to be renovated. The adopted approach responds to the need to integrate design and production through effective exchange of qualitative and quantitative information. The experiment tested the compatibility between acquired and processed data on a case study and the information structured in BIM libraries. The results include the application of parametric families to the building model and the validation of the resulting model for the intervention design. The possibility of exporting the geometric data in an open format allowed us to meet the information requirements for semi-automated production in the factory. However, limitations are highlighted due to the loss of parameterization resulting from the exchange format, which makes modification difficult during the pre-production phase. The approach, based on tailored yet flexible digital workflows, enabled the optimization of the entire process, contributing to a reduction in the risk of errors, work time, and improved communication between designers, manufacturers, and installers, overcoming traditional difficulties in the interaction between phases. [IT] La ricerca affronta il tema della modellazione digitale del costruito e dei componenti edilizi prefabbricati, esplorando i rapporti tra la rappresentazione del rilievo, l’analisi dello stato di fatto, la progettazione dell’intervento e la produzione. In particolare ci si è focalizzati sui differenti possibili usi dei modelli informativi per semplificare i flussi di lavoro, correlando due sperimentazioni: lo sviluppo di famiglie parametriche di pannelli prefabbricati usati per rivestire e migliorare le prestazioni dell’esistente; lo studio di strategie di Building Information Model dello stato di fatto di un edificio residenziale da riqualificare. L’approccio adottato risponde alla necessità di integrare progettazione e produzione attraverso uno scambio efficace di informazioni qualitative e quantitative. Le sperimentazioni verificano la compatibilità tra le informazioni strutturate nelle librerie BIM e i dati acquisiti ed elaborati su un caso di studio reale. I risultati comprendono l’applicazione delle famiglie parametriche al modello dell’edificio e la validazione del modello risultante per la progettazione dell’intervento. La possibilità di esportare i dati geometrici in formato aperto ha, in seguito, permesso di rispondere alle esigenze operative poste dalla produzione semi-automatica in stabilimento. Tuttavia, si evidenziano alcuni limiti di interoperabilità legati alla perdita di parametrizzazione nei processi di esportazione e importazione dei file, con conseguente difficoltà di modifica dei file stessi in fase di pre-produzione. L’approccio, basato su flussi di lavoro digitali appositamente studiati e flessibili, permette di ottimizzare l’intero processo riducendo i rischi di errore e i tempi di lavoro, migliorando la comunicazione tra progettisti, produttori e installatori e superando le note difficoltà di interazione tra le fasi.

Heritage masonry structures, with their complex geometries and variety of materials, pose significant challenges for accurate digital modelling and simulations for restoration and conservation purposes. In addition, traditional cultural heritage (CH) documentation workflows typically rely on geometric 3D acquisitions, often without supporting material diagnostics or structural insights. This work addresses this gap by introducing a data-driven workflow that leverages image-based information and 3D point clouds to extract both geometric and material-related attributes for energy analysis applications. The proposed methodology leverages Deep Learning (DL) and Finite Element Method (FEM) modelling to support energy simulation for cultural heritage assets. The proposed workflow integrates orthoimages and 3D data to segment masonry textures, estimate wall thickness, and generate a semantically enriched mesh tailored for energy analyses. A YOLO-based model identifies stone and mortar regions in high-resolution imagery, while point cloud voxelization and plane fitting are used to compute local thickness values. This information feeds into an adaptive meshing strategy, where mesh resolution is adjusted based on material texture and geometric features. A tunable parameter β enables control over mesh density, allowing for optimization of computational performance in thermal FEM simulations. This approach enables the derivation of meaningful simulation-ready 3D models from limited survey data.

Building renovation to improve energy efficiency is crucial for reducing CO₂ emissions, aligning with the goal of achieving net-zero emissions by 2050. This task requires a holistic approach that encompasses retrofitting outdated systems, enhancing thermal insulation, and integrating renewable energy sources. Simulating different indoor environmental conditions and technological systems within Digital Twin (DT) before interventions is crucial for optimizing energy efficiency. Simulations can support the proper installation of heating and cooling devices and facilitate the deployment of advanced technologies, including smart Heating, Ventilation, and Air Conditioning (HVAC) systems, energy-efficient lighting, and automated energy management solutions. The use of Artificial Intelligence (AI) in simulations allows for the precise sizing of HVAC systems, including heat pumps and related devices, by accurately modelling demand profiles and optimizing sensor placement based on the geometries of DTs. This study, conducted as part of the Horizon Europe InCUBE project1, explores a real-world use-case at the Centro Servizi Culturali Santa Chiara in Trento, Italy. It introduces an innovative approach that integrates 3D surveying, computational fluid dynamics (CFD), and digital twin (DT) geometries to enhance the analysis of indoor heat distribution. The proposed data-driven pipeline optimizes sensor placement within indoor spaces, ensuring precise system design, improving performance and energy efficiency, and minimizing energy waste while preventing the oversizing of technological systems.

The creation of Building Information Models (BIM) is driven by cutting-edge software applications, plug-ins, and APIs that constitute the backbone of BIM authoring tools. While free tools and APIs offer visualization and customization options, geometric modelling remains largely restricted to interactive work and proprietary platforms, which sometimes limits flexibility and efficiency. There are still only a few comprehensive workflows that fully automate the reconstruction of building elements from reality-based surveyed data. This paper introduces an innovative reconstruction pipeline developed for the Scan-to-BIM Challenge at the CVPR 2024 Workshop, where it achieved second place in the competition. A deep learning (DL)-driven BIM Module for parametric IFC reconstruction is designed to accurately reconstruct both primary and secondary building elements within a BIM framework, starting from unstructured point cloud data captured via Terrestrial Laser Scanning (TLS). By leveraging DL techniques, particularly Convolutional Neural Networks (CNNs) and Transformers Networks (PTv3), our approach uses late fusion instance segmentation across both 2D and 3D modalities to accurately identify and reconstruct class-specific elements. The pipeline ultimately generates Industry Foundation Classes (IFC) elements, enhancing modelling accuracy, parameter estimation, and consistency in subsequent stages. Results highlight the pipeline’s strong performance on various datasets, underscoring the crucial role of DL in advancing Scan-to-BIM workflows.

In many countries, recent boosts in the construction and renovation sectors and energy efficiency directives are driving a growing interest in the built environment among designers and maintainers. In this context, customized software solutions tailored for Building Information Modelling (BIM) and Building Energy Modelling (BEM) are proving to be indispensable for optimizing operational efficiency within the Architecture, Engineering, Construction, Owner, and Operator (AECOO) sector and for facilitating the generation of buildings' Digital Twins (DTs). These DTs rely on accurate geometry and ancillary information (semantics, sensors, etc.) to define properties of single elements, enabling crucial simulations in structural conditions or energy needs. However, BIM and BEM model creation and their enrichment with semantic information are highly time-consuming and prone to manual errors. Hence, there is an increasing demand for automatic methods featuring a high level of geometric accuracy to reconstruct building elements, such as walls, floors, and openings captured via 3D reality-based surveying. This paper introduces an automated method for creating Boundary Representation (B-Rep) models from 3D surveying data for the generation of digital building replicas. The method is based on the detection and computation of topological elements from 3D reality-based point clouds. It proves valuable for architectural or design workflows and for conducting energy or quality system simulations.