Architectural intelligence最新文献

Augmented reality (AR) technology has the potential to play a pivotal role in on-site construction activities by providing workers with hands-free access to virtual content and information through interactive experiences overlayed on the real world. The research presents a framework to improve design to production workflows in prefabricated timber building scenarios through the application of AR-driven assembly processes. An AR environment was developed to aid non-expert self-builders to a CLT based built scenario, understand the building assembly process, and efficiently conduct on-site assemblies. The AR environment, deployed through a head-mounted device (HMD), provides an elevated user experience through an interactive design interface, assembly sequence guide and module library. It aims to provide a communication line between the virtual components to the physical world by employing the use of continuous QR code marker-based tracking and virtual snap-to-place features of building components. A case study is presented for a CLT-based 64sqm family house in an urban village in China context to demonstrate the applications of an AR-assisted assembly. The framework was tested through a 1:5 scaled model prototype, which concentrated on flat timber-based components. User evaluation quantified the perception and experience of the builder on aspects such as constructability and operability of the model and user experience of the AR environment. The survey concluded that the AR-assisted assembly of the prototype demonstrated the potential for a self-builder to build with the guidance of AR technology. The outcomes of the research continue to push the boundaries of improved assembly tasks and provide evidence-based research on the benefits of integrating AR technology in production workflows.

The growing population of people living with dementia demands innovative architectural solutions that prioritize wellbeing. Floor layouts significantly affect the quality of life for people living with dementia, but the nuances of perceived user experience remain a challenging criterion to evaluate during the early design stages. Visual sightlines are one of the key aspects for dementia-inclusive design where machine learning (ML) provides decision-support means to validate early-stage designs. In this study, an isovist-based quantification scheme is proposed to capture visual access data to evaluate the extent of compliance of floor layouts with respect to dementia design principles (DDP). The visual access quality labels are determined by the number of isovists that satisfy the visual access requirement for their respective DDPs, thereby ensuring a consistent and objective measurement of visual access. 18 unique spatial features were tested using feature filtering methods to predict visual access quality labels. The framework bridges qualitative design principles with quantitative methods, introducing a scalable approach for evaluating visual access in the context of dementia-friendly design. The ML model was evaluated using individual class output and multi-output evaluation metrics based on 7 feature inputs and 2 class outputs, achieving 84–87% accuracy on an individual class and 72% on the subset accuracy metric. The results suggest a viable pathway for developing ML support tools to provide feedback on DDP compliance at an early design stage.

Most AI-enhanced architecture design tasks are confined to individual use and linear design flows, with a steep learning curve for mastering complex AI tools. Although advanced AI systems have shown enormous potential for improving human-AI in creative collaboration, their use in architectural design has yet to be fully explored. This paper proposes a conversational design framework for early-stage architectural design ideation and exploration, featuring a WeChat-integrated, architecture-centric multimodal agentic AI system. Leveraging the common messaging application facilitates the medium for design collaboration. A case study of a design workshop for a mixed-use renovation development in Suzhou old town is presented. Participants included undergraduate architecture students, and the instructors joined a WeChat group with the developed AI bot for design conversation and collaboration. Each student developed a design proposal, and three modes of conversational design were observed: performance-based, evaluation-based, and designerly-based. The challenges, opportunities and implications of research, education, and real-world practice are discussed through observation, count of interactions, questionnaires and semi-structured interviews with the students.

The article interprets tectonics in the age of artificial intelligence with the concept of ‘Neural Tectonics.’ In tectonics, poetics are presented through the assembly of materials. In contrast, neural tectonics is the synthetic poetics presented through the structure of deep neural networks and the semantics of training datasets. Around the concepts of artificial/authentic and artifacts, the article touches on issues important to both architecture and computing, including representation and geometrical projection space/latent space, and elaborates this process with the work Neural Artefact Black. Neural Tectonics seeks a kind of AI-native architecture, which is based on the generative logic of AI rather than the imitation and consumption of architectural styles. It promotes a reflection on ‘deep artificiality.’

Green roofs (GRs) play an important role in urban sustainability initiatives, offering reduced building energy consumption and enhanced ecological performance through additional growing medium and vegetation layers. Nevertheless, these functional components introduce challenges such as extra structural load, greater slab thickness, and increased construction complexity. This research proposes a design-fabrication method for a thin-vaulted green roof prototype, featuring a compression-only surface and upstand ribs along unevenly distributed stress lines to achieve both lightweight properties and material efficiency compared to a conventional flat roof with GR systems. To realize such a non-standard and multifunctional structure, a hybrid formwork system combining 3D clay printed (3DCP) molds and a reusable, doubly curved foam base is proposed to handle geometric complexity and functional integration by incorporating stay-in-place formwork as the growing medium. An empirical construction experiment, conducted at a 1:5 scale, validates the proposed method with optimized fabrication procedures for a proposed application scenario in a community renovation project.

Two-dimensional spatial analysis and trajectory-based environmental behavior research are widely used methods for investigating pedestrian wayfinding behavior. However, they often struggle to balance efficiency and granularity. This study draws inspiration from serious games to develop an online VR-based wayfinding platform for behavioral research, aiming to enhance data collection and improve resolution. A virtual replica of Satellite Terminal 1 (S1) of Shanghai Pudong International Airport (PVG) was created, providing participants with an immersive and intuitive environment. Accessible via any compatible web browser, the platform utilizes a browser/server (B/S) framework, significantly reducing the cost and complexity of behavioral data acquisition. Wayfinding tasks were designed with competitive mechanisms and reward-based elements to encourage public participation and exploration. Within three days of the platform's launch, 370 participants completed wayfinding tasks, generating a rich dataset of 2,746 trajectories and observed behaviors with spatial accuracy at the decimeter level and temporal accuracy at the second level. This study demonstrates that a gamified VR-based platform is a viable approach for researching navigation behavior, offering notable advantages in cost-effectiveness, participant engagement, and data acquisition efficiency.

Enhanced interactive AR bricklaying aims to combine human intelligence with robotic precision, enhancing creativity while addressing the inherent situation that the conventional robotic construction process are not well adapted to uncertainty. In conventional interactive brick-laying workflows, the design and construction phases are separate. Augmented Reality (AR) assists humans in completing parametric form designs, after which robots precisely execute the bricklaying process. However, this method demonstrates a low level of human–robot interaction, failing to fully harness the potential of AR in enhancing interactivity. This study first conducts a comparative experiment to evaluate a more interactive workflow, where design and construction phases alternate, against the conventional workflow. It examines the impact, limitations, and potential of human involvement in a linear robotic construction process. Based on this analysis, a construction case uses an AR system with a higher degree of interactivity. The system redistributes tasks between humans and robots, informed by the findings of the comparative experiment. The new workflow facilitates the construction of a brick wall, composed of large engineering bricks without mortar, measuring 7 m by 1.8 m, and featuring an inclined pattern. AR-guided interaction enables humans to guide robotic construction of a simple structure without relying on simulations or obstacle avoidance path planning. The discrepancies observed between design and final construction outcomes highlight the role of human intelligence, further validating the significance of enhanced interactive construction as a promising approach.

Modern campuses integrate various urban infrastructures such as residential, transportation, commercial, and service facilities. Research on personnel behavior within campuses provides valuable reference for campus management, resource allocation, and spatial planning. The integration of multi-source data enables a more comprehensive and nuanced analysis of the interplay between human behavior and spatial distribution, offering deeper and more objective insights into these phenomena. This study focuses on the Wuxi campus of Southeast University, combining Wi-Fi probe localization technology with a custom-developed physical environment sensor system. Based on the pre-experiment monitoring of pedestrian flows, high-traffic public spaces were selected for analysis. Over a period of 14 days, 28.87 million raw localization data points and 340,000 raw environmental data points were collected. To ensure the accuracy and reliability of the data, rigorous data cleaning procedures were employed, including the time restrictions, anomalous data, drift, redundant and ping-pong effects. Both the localization data and the physical environment data were subsequently visualized, allowing for the detailed examination of temporal and spatial behavioral characteristics of campus occupants and the corresponding environmental attributes of public spaces. By fitting the two heterogeneous datasets, the study revealed a correlation between the spatial distribution of pedestrian stopping behavior and physical environmental factors. Additionally, a convolutional autoencoder neural network was used to extract both 2D and 3D features of pedestrian trajectories, yielding four typical spatiotemporal patterns of pedestrian movement. These findings provide a robust foundation for improving the physical environment of the campus and informing crowd management strategies, while also offering critical insights for future campus planning and the optimization of spatial resources.

The growing demand for inhabitable spaces drives increased reliance on energy-intensive construction materials such as concrete and steel, which significantly contribute to global carbon emissions and resource depletion, highlighting the urgent need for sustainable alternatives. Renewable, bio-based materials like timber provide viable solutions, offering carbon sequestration and reduced environmental impact but lead to challenges related to biodiversity conservation, land use, and sustainable forest management. Natural fibers such as flax are increasingly used in sustainable composite materials and exhibit short growth cycles, minimal environmental impact, and favorable mechanical properties. When combined with timber, natural fiber-timber hybrids offer a large potential for high-performance, resource-efficient structural building parts. By leveraging the complementary strengths of both materials, such hybrids reduce reliance on valuable timber resources, replacing them with fast-growing flax fibers. To realize this potential for natural fiber-timber hybrid beam elements, existing design, evaluation, and fabrication methods for fibrous building parts are expanded and adapted. Suitable material candidates for fiber-timber hybrids are classified and characterized, and morphological parameters are defined to design and evaluate novel beam typologies. To allow for the manufacturing of large-scale natural fiber bodies for use in hybrid beam elements, new robotic fabrication methods are introduced, and existing manufacturing equipment is expanded. These innovations are exemplified by the Hybrid Flax Pavilion, the first permanent building to incorporate load-bearing natural fiber-timber hybrid components.