The construction industry lags behind other industrial sectors in terms of digitalization. The construction sector's slow adoption of comprehensive digital transformation is due to significant fragmentation throughout the building lifecycle and numerous stakeholders with diverse interests and priorities, hindering effective coordination and thus delaying innovative technology implementation. Following a design science approach, we propose and develop a process-oriented analysis framework for servitization based on service-dominant logic and Gaia-X-based ecosystem standards and technologies. The servitization approach is illustrated by applying it to the building permit process in general and described with exemplary services. The novelty of the study lies in combining service-dominant logic with a process-oriented, design-science framework that is operationalised via Gaia-X ecosystem standards and technologies to systematically derive digital service opportunities along the building lifecycle.
While a lot of research has examined semantics and digital twinning (DT), their use in bridge network monitoring has received less attention. In particular, the lifecycle management of information on bridge infrastructure is a critical challenge, and this paper aims to address it through a linked data framework that meets current needs and aligns with common practices in bridge maintenance. The paper proposes a technological framework based on knowledge graphs, open linked data ontologies, and industry standards such as IFC, INSPIRE, and ICDD to achieve multi-scale data integration, interoperability and standard data delivery. This framework serves as the baseline architecture for a knowledge graph-based DT system. It uses a microservices approach, assembling a suite of open-source Extract, Transform and Load (ETL) components, a graph database, and a file database. The application on a bridge network in the Metropolitan Area of Barcelona demonstrates that the proposed framework automatically abstracts Bridge Condition Indices (BCI) at the network level by aggregating state updates from individual bridge components. Where official inspection scoring procedures were unavailable, these indices utilise heuristic approximations to illustrate the prototype workflow. Furthermore, the system enables the resolution of context-rich queries that combine GIS, BIM, and inspection data. This study demonstrates how Digital Twins can be conceptualised and developed to operate with standardised, flexible, modular, interoperable, and multi-scale data, setting the groundwork for future Bridge Management Systems (BMS).
Due to the dynamic nature of construction sites, constant installation and removal of safety equipment is a required practice. This to date requires manual, thus infrequent and still time-consuming inspections to ensure the safety measures are correctly in place when needed. This paper introduces a novel end-to-end pipeline that integrates SafeConAI, UAV-collected point cloud data, and 4D BIM to automate safety inspections, overcoming the limitations of fragmented prior approaches by enabling real-time deviation mapping and comprehensive safety monitoring. This approach aims to bridge the gap between as-built data integration with autonomous navigation by providing location-aware compliance checking. Autonomous ground or aerial vehicles collect point clouds, and a segmentation model is trained to detect and segment safety guardrails and other essential building elements. The system is built and tested in a laboratory environment first by creating a typical fall-from-height protection case. The information generated from the point cloud segmentation is then compared with the original BIM to monitor and point out deviations, and finally provides an analysis report in a user-friendly SafeBIM. The method has been evaluated on several test cases from laboratory to real construction site settings. The results from this work indicate a promising method that can assist practitioners in (semi-) automating construction site safety inspection and making workplaces safer.
Safety training in high-risk industries often employs immersive technologies to raise risk awareness, yet this does not always translate into evidence-based effective safe practice. Battery Energy Storage Systems (BESS) are increasingly deployed across construction and energy infrastructure, but training personnel to operate and maintain them safely remains challenging due to high risks. Immersive Virtual Reality (VR) offers a promising solution by enabling realistic, hands-on practice of hazardous procedures in a controlled environment. This study addresses whether VR's immersive engagement translates into improved procedural learning outcomes for a safety-critical BESS soft-shutdown procedure. This is done by evaluating an immersive VR training module against conventional video-based instruction in a randomized experiment with 60 engineering students. Results show that VR training significantly improved spatial understanding of equipment layout (27% higher scores) and procedural comprehension (28% higher), while VR participants completed productive tasks 45% faster on average. Participants also rated VR as more engaging with greater learning confidence, though additional instructor guidance was often needed for navigation and interaction. However, a confidence-competence mismatch was observed with perceived abilities exceeding objective performance in some domains. This motivates embedded assessment and feedback. While both training methods supported similar recall of step sequences, neither adequately conveyed the underlying safety rationale behind the procedure. These findings suggest that while VR enhances certain learning outcomes through interactive, risk-free practice, it is not a standalone solution for full procedural mastery. Practical implications include design guidelines for VR training (e.g., explicit rationale cues and embedded assessments) and recommendations for integrating VR with construction safety programs to strengthen workforce readiness for emerging technologies in the built environment.
Computational Design (CD) has emerged as a key digital technology in the Architecture, Engineering and Construction (AEC) industry, enabling designers to automate tasks, explore complex design alternatives, and collaborate more effectively. Despite these advantages, the adoption of CD faces numerous challenges related to limited knowledge, planning, change management, code readability, and data management. This study investigates how these challenges can be mitigated through the Methodology of Knowledge-Based Engineering Applications (MOKA). Using previously identified CD challenges from Mikaelsson (2022), each challenge was categorized according to the six phases of MOKA—Identify, Justify, Capture, Formalize, Package, and Activate—to analyse where mitigation can occur. The results show that most challenges can be addressed within the first two phases, Identify and Justify. Furthermore, MOKA complements existing research by offering a structured, process-oriented framework to guide CD development and implementation. Although MOKA has its limitations, the study concludes that there are opportunities to enhance computational design practice in AEC if parts of MOKA is applied iteratively during building design. Further details on our mitigation approach are presented and discussed in this study.
Effective worker health and safety (H&S) management remains a major concern in the construction industry. Traditional safety measurement methods, while valuable, fall short of achieving zero injuries in dynamic and complex work settings. In response to this limitation, researchers have turned to emerging techniques, such as artificial intelligence (AI), to enhance the efficiency of H&S management methods and develop predictive models for on-site mitigation of occupational hazards. Despite numerous studies showcasing potential benefits and limitations of AI in various construction H&S applications, a comprehensive synthesis tailored to construction workers' H&S management is needed. This study conducts a bibliometric and systematic literature review of various AI strategies (specifically machine learning and deep learning) for ensuring effective H&S management and identifies gaps in existing research. Leveraging a systematic review approach, 181 articles from relevant academic journals and conferences published up to July 2025 were analyzed. The findings suggest a rapid increase in AI-related H&S research, particularly in Asia and North America. Results revealed seven construction H&S management AI application themes: construction accidents, PPE detection, ergonomic risk, safety inspections, fatigue, safety behavior, and health and safety training. The study highlights the strengths and weaknesses of current applications and proposes areas for further investigation. This review offers foundational insights essential for developing robust prediction models and advancing the use of safe and ethical AI in construction H&S management.
Construction scheduling still relies on fragmented manual preparation of activity, location, and labor data, even when BIMmodels are available. This paper addresses the challenge ofsystematically structuring and automatically extracting enriched BIM data to enable schedule-ready inputs for advanced construction scheduling. The paper develops an Enhanced Planning and Scheduling (EPS) based framework and applies it through BIM enrichment, spatial decomposition, classification mapping, labor-hour calculation, and Dynamo-based data extraction in a building case. The framework generates a reusable dataset that links model elements to zones, floors, workspaces, activity categories, labor hours, and EPS priority IDs for downstream scheduling. The papers' contribution lies in providing a structured EPS-based framework that transforms enriched BIM data into reusable schedule-ready inputs linking locations, activity categories, labor hours, and priority logic for downstream scheduling. Further research is needed to validate the framework across various project types and scheduling workflows, and to examine dynamic model-schedule updating andAI-assisted scheduling integration.
This study addresses a persistent challenge in computer vision for construction monitoring: deep learning models trained on source-domain data often perform poorly when deployed in new target domains due to distribution shifts and limited annotations. To mitigate these issues, the research introduces TDG-CIS, a clustering-initialized semi-supervised framework designed to generate high-quality instance segmentation training data directly from unlabeled target-domain images. TDG-CIS operates in two stages. First, it employs a clustering-based mask generation strategy that uses a transformer feature backbone to extract patch-level representations and derive initial instance masks without human supervision. These masks serve as a reliable starting point for semi-supervised learning. Second, a semi-supervised instance segmentation model iteratively refines these masks and converts raw images into usable training samples. This iterative pipeline allows the model to progressively improve segmentation quality while adapting to the visual characteristics of diverse construction environments. The framework was validated on a large dataset of 50,000 images spanning more than 70 construction-related domains. Experimental results show that TDG-CIS achieves a 77.9% data utilization rate, along with 87.5% mAP and 81.1% mAR in segmentation quality. When used to scale training data for downstream instance segmentation models, TDG-CIS yields substantial performance gains: baseline models trained on automatically generated data outperform those trained on manually labeled datasets, improving mAP from 92.9% to 94.3% and mAR from 86.7% to 88.6%. Ablation studies further demonstrate that the semi-supervised refinement mechanism is key to boosting both data utilization and segmentation accuracy. Overall, the study offers a novel approach that eliminates dependence on source-domain supervision and provides a scalable pathway for producing target-domain training datasets for instance segmentation in intelligent construction applications.
Building Information Modeling (BIM) environments contain structured, data-rich models, yet the design logic embedded within them is rarely reused beyond individual projects. Although BIM systems capture relationships between spatial configurations and building components, this knowledge typically remains project-specific and is not systematically utilized in future design processes. This study introduces a machine learning (ML)-based computational framework that predicts door specification attributes from room-to-room spatial transitions, aiming to support data-informed decision-making in design. In this context, BIM is approached not only as a modeling environment but also as a source of transferable design knowledge. A dataset of 5,763 door instances was compiled from three Revit-based projects representing different building typologies: a hospital, an office, and a recreational facility. Spatial transitions were encoded using a Bag-of-Words (BoW) representation, and Random Forest algorithms were applied for classification and regression. Four scenarios were tested: (i) intra-project learning, (ii) cross-project learning, (iii) combined project learning, and (iv) domain-augmented retraining. To provide a benchmark, model performance was evaluated against simple baseline strategies, including majority-class prediction for classification and mean-value prediction for regression. Results showed strong performance for some targets within single projects, while others exhibited more moderate results. Performance decreased across projects due to inconsistent naming, class imbalance, and label mismatch, where the model encounters previously unseen categories. The findings demonstrate that archived BIM models can be transformed into predictive design intelligence, enhancing computational reasoning and efficiency within BIM-based workflows, while emphasizing the need for data standardization to achieve robust cross-project applicability.
Tower cranes are critical for lifting heavy elements in prefabricated and modular construction, but their operations pose significant safety risks. This study proposes a data-driven method using Real-Time Kinematic Global Navigation Satellite System (RTK-GNSS) that tracks the crane trolley and workers movements and assesses the potential struck-by hazards from lifted payloads. By analyzing the trolley’s velocity and vertical displacement, the novelty in the method respectively detects the lifting stages and estimates the payload weight, while pairing its real to the planned placement location in the Industry Foundation Classes (IFC) model. An energy-based hazard assessment computes the intensity of detected incidents when workers’ RTK-GNSS wearables are inside of hazardous crane swing zones. The severity is evaluated in the form of a statistical analysis and density map. Unprecedented safety-relevant information becomes available to practitioners that can use it in responsible decision-making. Compared to computationally intensive systems such as cameras, the proposed method provides an alternative cost-effective, scalable solution for automating the monitoring of hazardous outdoor workspaces.
Construction relies on rapid, reliable information sharing among distributed teams to enable project success. This article presents an empirical exploration of eleven organizations that serve as early adopters of advanced information technologies for instantaneous, i.e., real-time or near real-time, project controls and reporting capabilities. The study employed a grounded theory approach, in which analyses informed subsequent steps. Constructs were coded from the transcription of interviews with corporate managers affiliated with owner, contractor, software developer, and manufacturing organizations. Contextual and processual findings were leveraged to support a theoretical model that captured the changes and implementation steps for successful technology diffusion. The study also defined the drivers that demand and justify an instantaneous or frequent reporting capability in the design and construction sector. Results offer actionable insights. Organizations can more effectively adopt advanced project controls by proactively planning and embracing necessary organizational changes. Researchers should consider both contextual and processual factors to support effective exploration of technology integration in construction.
This study assesses transformer-based 3D semantic segmentation models for detecting structural components in terrestrial laser scans, given that no training data currently exists for shell construction sites. Manual annotation of 3D point clouds is expensive, yet high-quality labels remain essential for supervised computer vision and validation. Automated pre-labeling can cut down annotation effort by shifting human tasks from exhaustive labeling to targeted verification and correction, assuming models can robustly identify the most common structural elements. We designed a three-stage evaluation protocol covering (i) supervised learning, (ii) cross-domain generalization, and (iii) transfer learning with limited labeled data in the target domain to test model generalization in this context. Three transformer architectures (Point Transformer V2, Point Transformer V3, and Swin3D) are evaluated using four established indoor datasets (S3DIS, ScanNetV2, Structured3D, and VASAD) and a custom domain-specific dataset of annotated construction scenes. Training only on the limited construction dataset results in weak generalization. In contrast, pretraining on loosely related synthetic data and fine-tuning on a minimal number of labeled construction scenes enable reliable segmentation of core building components. A sensitivity analysis also showed that just 12 samples are sufficient to calibrate a pretrained model to a specific building type. The models perform well despite differences between synthetic training data and noisy real-world scans. Among the evaluated architectures, Swin3D delivers the best performance, with +18% mIoU improvement through general pretraining, while PTv3 converges faster with fewer target-domain samples. These findings suggest that transfer learning with limited labeled construction data offers a practical foundation for scalable pre-labeling workflows and human-in-the-loop applications in architecture, engineering, and construction.
As part of Industry 4.0 initiatives, the construction industry is increasingly adopting Digital Twin (DT) to enhance asset lifecycle management, predictive maintenance, and data-driven decision-making. However, DT implementation remains fragmented and uneven across lifecycle phases, application domains, and organisational contexts. This study aims to address these gaps through a comprehensive review of current DT practices in construction. A two-stage systematic literature review, following PRISMA guidelines, was conducted. The first stage analysed 122 DT review articles to map thematic trends, research focuses, and overlooked areas. The second stage synthesised 297 empirical studies to examine practical application distribution, technology integration frameworks, deployment barriers, and mitigation strategies. Current DT research is heavily concentrated on the operation and maintenance phase, with limited attention to early design or end-of-life activities. Key challenges include data fragmentation, interoperability issues, high initial costs, limited stakeholder engagement, and insufficient regulatory and organisational support. A range of technical and institutional strategies has been identified to address these barriers. Crucially, the study translates these findings into actionable roadmaps for key stakeholders, offering role-specific strategies to bridge the gap between theory and practice. This study presents a comprehensive synthesis of over 400 publications from 2019 to 2024, systematically mapping DT applications across lifecycle stages, categorising key barriers, and evaluating targeted strategies for each. By identifying critical knowledge gaps and limitations within the current body of DT research, it offers valuable insights to inform future investigations and support more scalable and integrated implementation in practice.
As exoskeletons gain traction in the construction industry, evaluating the ethical and social dimensions of exoskeletons and devising strategies to mitigate these risks becomes imperative. This review focuses on assessing the ethical and social risks associated with the integration of exoskeleton technology in construction, with a goal to enhance worker safety and well-being. Exploring both the potential benefits and challenges of exoskeleton usage, the paper underscores the importance of a balanced approach that reconciles technological advantages with ethical considerations. A systematic literature review was conducted to gather insights into the ethical and social aspects of incorporating exoskeletons in the construction industry. The research involved a comprehensive analysis of existing literature. While the study’s background provides a comprehensive overview of the current state of exoskeleton usage in the global construction industry, this review reveals significant ethical and social concerns surrounding exoskeletons in construction. These include device design, stigmatization, regulatory standards, worker consent and autonomy, trust, potential job displacement, and data privacy. Social considerations include accessibility and affordability, human rights, cultural diversity, and social communication. Effectively addressing these risks requires the establishment of clear ethical guidelines, training, vigilant monitoring, compliance, public engagement, government intervention, and collaboration with researchers and industry stakeholders. While exoskeletons hold the potential to reduce musculoskeletal disorders and ergonomic risks, addressing ethical and social risks is paramount. Neglecting these aspects may impede the acceptance and adoption of exoskeletons, leading to risks such as misuse, decreased social communication, and job displacement. The study proposes a framework that offers insights for industry stakeholders and guides the ethical adoption of exoskeleton technology. A collective effort is necessary to ensure the responsible integration of exoskeletons, fostering a safer and more sustainable construction industry and optimizing their advantages while mitigating disparities and discrimination in the construction industry.
While many regions have accelerated Building Information Modelling (BIM) through government mandates, many jurisdictions must pursue adoption without centralised enforcement. This raises a key question: what governance arrangements enable consistent BIM information management andISO 19650 alignment in non-mandated settings? This paper addresses this question through a longitudinal case study of Wales, a devolved UK region without a devolved BIM mandate, programmatic funding, or a coordinated implementation framework. Using a qualitative-dominant mixed-methods design comprising stakeholder workshops (2017-2023), semi-structured interviews, and a structured survey of contractors and local authorities, the study examines how stakeholders interpret and operationalise ISO 19650 in practice. The results indicate that reported BIM tool use is widespread, yet formal standards-aligned implementation is substantially lower, revealing a persistent gap between software adoption and ISO 19650-oriented information management. Across data sources, recurring governance constraints include weak and inconsistent client-side commissioning of information requirements, project documentation that is produced primarily for compliance rather than embedded coordination, and ambiguity in information management roles and accountabilities across the project lifecycle. Rather than treating mandates as a prerequisite, the study argues that governance maturity is a primary enabling condition for ISO 19650 alignment in non-mandated contexts. The paper contributes empirical evidence on bottom-up BIM governance and proposes practical, non-legislative interventions, including shared regional templates, a knowledge hub to translate requirements into practice, role-specific training emphasising applied competence, and digital support for validation and assurance, measures intended to reduce interpretive burden and improve consistency across the supply chain.
Building renovation is essential to reduce environmental impacts and address social demands, yet structural rehabilitation planning remains uncertain and disruptive. This paper presents Endurify 2.0, a BIM-based automated scheduling system for structural maintenance that estimates the Remaining Useful Life (RUL) of reinforced concrete beams and automatically generates multi-phase rehabilitation plans. By minimizing manual planning input and integrating maintenance data into the BIM model, the system serves as a decision-support tool that enhances the efficiency and accuracy of rehabilitation planning. The tool combines analytical models for four damage indicators to define element-level intervention thresholds. These outputs feed a multi-criteria decision-making framework, economic cost, and two social criteria are weighted, and TOPSIS selects the optimal rehabilitation schedule. The method is validated on a residential building with 191 beams, 139 exhibiting damage. Compared to expert-based planning, the automated approach reduces total cost by 15% and proximity impact by 10%, while maintaining structural safety.
This paper presents an automated framework for generating high-fidelity bridge information models based on the IFC4x3 standard. Addressing the challenges of manual modeling and data consistency in bridge engineering, the proposed solution enables the seamless transformation of structured design data (e.g., Excel tables) as input into detailed and semantically rich IFC models as output through programmatic generation. The workflow integrates a JSON intermediary layer to facilitate flexible data exchange and supports the rapid assembly of complex bridge components, including main girders, secondary beams, connections, bolts, stiffeners, and diaphragms. A key innovation lies in the systems' parameter-driven geometry generation, which allows for efficient adjustment and iteration of bridge designs. The framework ensures both geometric precision and semantic completeness, providing the geometric and semantic foundation for downstream workflows including structural modeling, asset management, and maintenance planning. Furthermore, the architecture is designed with future AI integration in mind, enabling large language models to interact with and modify bridge parameters via natural language commands. Case studies on steel plate girder and box girder bridges demonstrate the systems' capability to handle intricate structural details and generate models swiftly, with performance scaling linearly with complexity. While current limitations include a focus on steel structures and reliance on comprehensive metadata, the paper outlines future directions such as expanding to other bridge types, implementing automated design rule checks, and enhancingAI-driven design support. Overall, this research advances the digitalization and automation of bridge modeling, providing a robustfoundation for intelligent design, analysis, and lifecycle management within the civil engineering domain.
Advancements in immersive virtual reality (VR) raise new possibilities for multidisciplinary architecture, engineering, and construction (AEC) team collaboration; however, the empirical basis for assessing its effectiveness remains insufficiently examined. This paper investigated whether avatar body language and 3D markup tools in a head-mounted display (HMD) VR platform improve remote AEC team performance compared to a Building Information Modeling (BIM) platform configured with equivalent navigation and communication features. A controlled counterbalanced experiment was conducted with ten teams, using a mixed-methods evaluation framework that incorporated meeting duration, decision-reporting accuracy, participant feedback, and observational analysis. The results showed that immersive VR significantly reduced average meeting duration by 26.7% and improved final decision-reporting accuracy compared to the BIM platform, while qualitative findings indicated that avatar embodiment and visual collaboration tools supported more effective coordination. These findings inform technology adoption decisions for AEC practitioners and advance the empirical foundation for researchers evaluating immersive VR as a coordination platform for distributed project teams. They further motivate future research aimed at refining immersive collaboration features and advancing their systematic evaluation to support evidence-based integration into AEC practice.
Data exchange in the AECOO industry remains constrained by the closed architectures ofBIMtools and the hidden dependency of open standards such as IFC on incompatible geometric kernels. This paper asks whether kernel-independent, mesh-based formats specifically glTF and USD can address the interoperability, scalability and data-access limitations inherent in current kernel-dependent BIMparadigms. A qualitative, review-critical analysis was conducted, combining a comparative assessment of data standards (IFC, glTF, USD), technical documentation ofgeometric kernels and platforms, and case-based evidence from commercial and R&D projects. The analysis shows that mesh-based formats, combined with structured non-geometric data stored in formats such as JSON, CSV or SQL, offer a plausible and increasingly practical alternative to kernel-dependent workflows for a significant range ofAECOO use cases, particularly those involving visualisation, simulation, robotics andAI. These findings are relevant to BIM researchers, software developers, andAECOO practitioners seeking to reduce vendor dependency and build scalable, AI-compatible data environments. Future research should empirically validate the performance of granular, mesh-based data systems across diverse project scales and disciplines, and investigate standardised methods for converting parametric data to mesh representations without critical loss of fidelity.
Facilities Management (FM) is undergoing a rapid transformation driven by the adoption of IoT devices, building management systems, and building information models. This disruptive shift introduces significant cybersecurity threats, posing risks to safety, data privacy, and operational continuity. This paper investigates which specific configurations of organizational, technological, and human factors lead to cybersecurity breaches within FM environments. Moreover, there is a notable gap within the FM literature in terms of comprehensive understanding and strategic readiness regarding cybersecurity threats. To address this gap, this paper presents findings from an extensive survey involving 114 FM professionals who experienced cybersecurity breaches. A Fuzzy-set Qualitative Comparative Analysis (fsQCA) was utilized to identify ten distinct pathways and combinations of organizational, technological, and human factors that commonly lead to cybersecurity incidents. The analysis revealed ten distinct configurations where limited internal preparedness, financial constraints, and insufficient awareness converge to create sufficient conditions for a breach. These findings provide FM practitioners and security officers with a diagnostic taxonomy of "vulnerability profiles," allowing them to prioritize interventions based on their specific organizational constraints. This research establishes a foundation for longitudinal studies to test how these breach configurations evolve as FM systems become increasingly autonomous and
