Cumincad : CUMINCAD Papers : Search Results

CumInCAD is a Cumulative Index about publications in Computer Aided Architectural Design
supported by the sibling associations ACADIA, CAADRIA, eCAADe, SIGraDi, ASCAAD and CAAD futures

PDF papers
References

Hits 1 to 20 of 652

Reformat results as: short short into frame detailed detailed into frame

_id	acadia20_688
id	acadia20_688
authors	del Campo, Matias; Carlson, Alexandra; Manninger, Sandra
year	2020
title	3D Graph Convolutional Neural Networks in Architecture Design
source	ACADIA 2020: Distributed Proximities / Volume I: Technical Papers [Proceedings of the 40th Annual Conference of the Association of Computer Aided Design in Architecture (ACADIA) ISBN 978-0-578-95213-0]. Online and Global. 24-30 October 2020. edited by B. Slocum, V. Ago, S. Doyle, A. Marcus, M. Yablonina, and M. del Campo. 688-696.
doi	https://doi.org/10.52842/conf.acadia.2020.1.688
summary	The nature of the architectural design process can be described along the lines of the following representational devices: the plan and the model. Plans can be considered one of the oldest methods to represent spatial and aesthetic information in an abstract, 2D space. However, to be used in the design process of 3D architectural solutions, these representations are inherently limited by the loss of rich information that occurs when compressing the three-dimensional world into a two-dimensional representation. During the first Digital Turn (Carpo 2013), the sheer amount and availability of models increased dramatically, as it became viable to create vast amounts of model variations to explore project alternatives among a much larger range of different physical and creative dimensions. 3D models show how the design object appears in real life, and can include a wider array of object information that is more easily understandable by nonexperts, as exemplified in techniques such as building information modeling and parametric modeling. Therefore, the ground condition of this paper considers that the inherent nature of architectural design and sensibility lies in the negotiation of 3D space coupled with the organization of voids and spatial components resulting in spatial sequences based on programmatic relationships, resulting in an assemblage (DeLanda 2016). These conditions constitute objects representing a material culture (the built environment) embedded in a symbolic and aesthetic culture (DeLanda 2016) that is created by the designer and captures their sensibilities.
series	ACADIA
type	paper
email
last changed	2023/10/22 12:06

_id	acadia20_238
id	acadia20_238
authors	Zhang, Hang
year	2020
title	Text-to-Form
source	ACADIA 2020: Distributed Proximities / Volume I: Technical Papers [Proceedings of the 40th Annual Conference of the Association of Computer Aided Design in Architecture (ACADIA) ISBN 978-0-578-95213-0]. Online and Global. 24-30 October 2020. edited by B. Slocum, V. Ago, S. Doyle, A. Marcus, M. Yablonina, and M. del Campo. 238-247.
doi	https://doi.org/10.52842/conf.acadia.2020.1.238
summary	Traditionally, architects express their thoughts on the design of 3D architectural forms via perspective renderings and standardized 2D drawings. However, as architectural design is always multidimensional and intricate, it is difficult to make others understand the design intention, concrete form, and even spatial layout through simple language descriptions. Benefiting from the fast development of machine learning, especially natural language processing and convolutional neural networks, this paper proposes a Linguistics-based Architectural Form Generative Model (LAFGM) that could be trained to make 3D architectural form predictions based simply on language input. Several related works exist that focus on learning text-to-image generation, while others have taken a further step by generating simple shapes from the descriptions. However, the text parsing and output of these works still remain either at the 2D stage or confined to a single geometry. On the basis of these works, this paper used both Stanford Scene Graph Parser (Sebastian et al. 2015) and graph convolutional networks (Kipf and Welling 2016) to compile the analytic semantic structure for the input texts, then generated the 3D architectural form expressed by the language descriptions, which is also aided by several optimization algorithms. To a certain extent, the training results approached the 3D form intended in the textual description, not only indicating the tremendous potential of LAFGM from linguistic input to 3D architectural form, but also innovating design expression and communication regarding 3D spatial information.
series	ACADIA
type	paper
email
last changed	2023/10/22 12:06

_id	ijac202018404
id	ijac202018404
authors	Paul Nicholas, Gabriella Rossi, Ella Williams, Michael Bennett and Tim Schork
year	2020
title	Integrating real-time multi-resolution scanning and machine learning for Conformal Robotic 3D Printing in Architecture
source	International Journal of Architectural Computing vol. 18 - no. 4, 371–384
summary	Robotic 3D printing applications are rapidly growing in architecture, where they enable the introduction of new materials and bespoke geometries. However, current approaches remain limited to printing on top of a flat build bed. This limits robotic 3D printing’s impact as a sustainable technology: opportunities to customize or enhance existing elements, or to utilize complex material behaviour are missed. This paper addresses the potentials of conformal 3D printing and presents a novel and robust workflow for printing onto unknown and arbitrarily shaped 3D substrates. The workflow combines dual-resolution Robotic Scanning, Neural Network prediction and printing of PETG plastic. This integrated approach offers the advantage of responding directly to unknown geometries through automated performance design customization. This paper firstly contextualizes the work within the current state of the art of conformal printing. We then describe our methodology and the design experiment we have used to test it. We lastly describe the key findings, potentials and limitations of the work, as well as the next steps in this research.
keywords	Conformal printing, robotic fabrication, 3D scanning, neural networks, industry 4.0
series	journal
email
last changed	2021/06/03 23:29

_id	ecaade2020_047
id	ecaade2020_047
authors	Brown, Lachlan, Yip, Michael, Gardner, Nicole, Haeusler, M. Hank, Khean, Nariddh, Zavoleas, Yannis and Ramos, Cristina
year	2020
title	Drawing Recognition - Integrating Machine Learning Systems into Architectural Design Workflows
source	Werner, L and Koering, D (eds.), Anthropologic: Architecture and Fabrication in the cognitive age - Proceedings of the 38th eCAADe Conference - Volume 2, TU Berlin, Berlin, Germany, 16-18 September 2020, pp. 289-298
doi	https://doi.org/10.52842/conf.ecaade.2020.2.289
summary	Machine Learning (ML) has valuable applications that are yet to be proliferated in the AEC industry. Yet, ML offers arguably significant new ways to produce and assist design. However, ML tools are too often out of the reach of designers, severely limiting opportunities to improve the methods by which designers design. To address this and to optimise the practices of designers, the research aims to create a ML tool that can be integrated into architectural design workflows. Thus, this research investigates how ML can be used to universally move BIM data across various design platforms through the development of a convolutional neural network (CNN) for the recognition and labelling of rooms within floor plan images of multi-residential apartments. The effects of this computation and thinking shift will have meaningful impacts on future practices enveloping all major aspects of our built environment from designing, to construction to management.
keywords	machine learning; convolutional neural networks; labelling and classification; design recognition
series	eCAADe
email
last changed	2022/06/07 07:54

_id	ecaade2020_283
id	ecaade2020_283
authors	Sebestyen, Adam and Tyc, Jakub
year	2020
title	Machine Learning Methods in Energy Simulations for Architects and Designers - The implementation of supervised machine learning in the context of the computational design process
source	Werner, L and Koering, D (eds.), Anthropologic: Architecture and Fabrication in the cognitive age - Proceedings of the 38th eCAADe Conference - Volume 1, TU Berlin, Berlin, Germany, 16-18 September 2020, pp. 613-622
doi	https://doi.org/10.52842/conf.ecaade.2020.1.613
summary	Application of Machine Learning (ML) in the field of architecture is a worthwhile topic to discuss in the context of digital architecture. Authors propose to extend this discussion, presenting an integrated ML pipeline built with the state-of-the-art data science tools. To investigate the affordances of such pipelines, an ML model being able to predict the environmental metrics of a generalized facade system is created. This approach is valid for arbitrary facades, as long as the proposed design could be discretized in the form analogous to the data generated for the ML model training. The presented experiment evaluates the precision of the sunlight hours and radiation values predictions, aiming at the application in the early design phases. Conducted investigation builds up on the knowledge embedded in the Grasshopper and Ladybug toolsets. Potential application of Convolutional Neural Networks and categorical datasets for classifications tasks to increase the precision of the ML models have been identified. Possibility to extend the approach beyond the workspace of Rhino and Grasshopper is suggested. Further research outlook, investigating the data pattern recognition capabilities in relation to the three-dimensional forms discretized as multidimensional arrays, is stated.
keywords	Machine Learning; Environmental Analysis; Parametric Design; Supervised Learning
series	eCAADe
email
last changed	2022/06/07 08:00

_id	sigradi2020_60
id	sigradi2020_60
authors	Asmar, Karen El; Sareen, Harpreet
year	2020
title	Machinic Interpolations: A GAN Pipeline for Integrating Lateral Thinking in Computational Tools of Architecture
source	SIGraDi 2020 [Proceedings of the 24th Conference of the Iberoamerican Society of Digital Graphics - ISSN: 2318-6968] Online Conference 18 - 20 November 2020, pp. 60-66
summary	In this paper, we discuss a new tool pipeline that aims to re-integrate lateral thinking strategies in computational tools of architecture. We present a 4-step AI-driven pipeline, based on Generative Adversarial Networks (GANs), that draws from the ability to access the latent space of a machine and use this space as a digital design environment. We demonstrate examples of navigating in this space using vector arithmetic and interpolations as a method to generate a series of images that are then translated to 3D voxel structures. Through a gallery of forms, we show how this series of techniques could result in unexpected spaces and outputs beyond what could be produced by human capability alone.
keywords	Latent space, GANs, Lateral thinking, Computational tools, Artificial intelligence
series	SIGraDi
email
last changed	2021/07/16 11:48

_id	ecaade2020_456
id	ecaade2020_456
authors	Farinea, Chiara, Awad, Lana, Dubor, Alex and El Atab, Mohamad
year	2020
title	Integrating biophotovoltaic and cyber-physical technologies into a 3D printed wall
source	Werner, L and Koering, D (eds.), Anthropologic: Architecture and Fabrication in the cognitive age - Proceedings of the 38th eCAADe Conference - Volume 2, TU Berlin, Berlin, Germany, 16-18 September 2020, pp. 463-472
doi	https://doi.org/10.52842/conf.ecaade.2020.2.463
summary	The research presented in this paper investigates the development of "3D printed ceramic green wall", a technological Nature Based Solution (NBS) aimed at regenerating urban areas by improving spatial quality and sustainability through clean and autonomous energy production. Building upon previous research, the challenge of this system is to adapt additive manufacturing processes of ceramic 3D printing with biophotovoltaic systems while simultaneously developing digital and cyber-physical frameworks to generate site and user responsive design and autonomous solutions that optimize system performance and energy generation. The paper explores the complex design negotiations between these drivers, focusing particularly on their performance optimization, and finally highlights the system potential as exemplified through a successful implementation of a 1:1 site responsive wall prototype.
keywords	Nature based solutions; biophotovoltaic systems; additive manufacturing; responsive design; cyber-physical networks; augmented reality
series	eCAADe
email
last changed	2022/06/07 07:55

_id	acadia20_150
id	acadia20_150
authors	Gaudilliere-Jami, Nadja
year	2020
title	AD Magazine
source	ACADIA 2020: Distributed Proximities / Volume I: Technical Papers [Proceedings of the 40th Annual Conference of the Association of Computer Aided Design in Architecture (ACADIA) ISBN 978-0-578-95213-0]. Online and Global. 24-30 October 2020. edited by B. Slocum, V. Ago, S. Doyle, A. Marcus, M. Yablonina, and M. del Campo. 150-159.
doi	https://doi.org/10.52842/conf.acadia.2020.1.150
summary	This paper aims to contribute to a history of computational design and to a historiography of the field by proposing a study of the development of sociotechnical networks of computation in architecture between 1965 and 2020 as shown in AD magazine. The research focuses on two aspects: (1) a methodological approach for the constitution of a comprehensive history of the field and the application of that methodology to a corpus of items published in AD, and (2) questions the relevance of the outlook into computational design as given by the magazine in comparison to a more comprehensive history taking into account other sources. First, the paper presents the history and the editorial line of AD, as well as its pertinence as a primary source. Second, a brief account of the history emerging from this research is given, with a focus on four different periods: pioneering research of the 1960s–1970s, emergence of 3D modeling tools and the procedural winter in the 1980s–1990s, constitution of a large-scale academic and professional network in the 2000s, and democratization of algorithmic design tools in the 2010s. Third, observations are made on editorial choices of the magazine and the biases of its account of computational research, with a special focus on the period 2000–2020, during which many issues have been dedicated to computational design themes, therefore making potential biases more visible. Despite the preponderance of specific topics, editors, and contributors, AD magazine provides an outlook into key concerns of the community at given times. The main biases identified, including a strong focus on the themes of biodesign and rationalization of practices, mirror the biases of the computational field itself, demonstrating the value of AD as an archive for the history of the field.
series	ACADIA
type	paper
email
last changed	2023/10/22 12:06

_id	acadia20_382
id	acadia20_382
authors	Hosmer, Tyson; Tigas, Panagiotis; Reeves, David; He, Ziming
year	2020
title	Spatial Assembly with Self-Play Reinforcement Learning
source	ACADIA 2020: Distributed Proximities / Volume I: Technical Papers [Proceedings of the 40th Annual Conference of the Association of Computer Aided Design in Architecture (ACADIA) ISBN 978-0-578-95213-0]. Online and Global. 24-30 October 2020. edited by B. Slocum, V. Ago, S. Doyle, A. Marcus, M. Yablonina, and M. del Campo. 382-393.
doi	https://doi.org/10.52842/conf.acadia.2020.1.382
summary	We present a framework to generate intelligent spatial assemblies from sets of digitally encoded spatial parts designed by the architect with embedded principles of prefabrication, assembly awareness, and reconfigurability. The methodology includes a bespoke constraint-solving algorithm for autonomously assembling 3D geometries into larger spatial compositions for the built environment. A series of graph-based analysis methods are applied to each assembly to extract performance metrics related to architectural space-making goals, including structural stability, material density, spatial segmentation, connectivity, and spatial distribution. Together with the constraint-based assembly algorithm and analysis methods, we have integrated a novel application of deep reinforcement (RL) learning for training the models to improve at matching the multiperformance goals established by the user through self-play. RL is applied to improve the selection and sequencing of parts while considering local and global objectives. The user’s design intent is embedded through the design of partial units of 3D space with embedded fabrication principles and their relational constraints over how they connect to each other and the quantifiable goals to drive the distribution of effective features. The methodology has been developed over three years through three case study projects called ArchiGo (2017–2018), NoMAS (2018–2019), and IRSILA (2019-2020). Each demonstrates the potential for buildings with reconfigurable and adaptive life cycles.
series	ACADIA
type	paper
email
last changed	2023/10/22 12:06

_id	caadria2020_384
id	caadria2020_384
authors	Patt, Trevor Ryan
year	2020
title	Spectral Clustering for Urban Networks
source	D. Holzer, W. Nakapan, A. Globa, I. Koh (eds.), RE: Anthropocene, Design in the Age of Humans - Proceedings of the 25th CAADRIA Conference - Volume 2, Chulalongkorn University, Bangkok, Thailand, 5-6 August 2020, pp. 91-100
doi	https://doi.org/10.52842/conf.caadria.2020.2.091
summary	As planetary urbanization accelerates, the significance of developing better methods for analyzing and making sense of complex urban networks also increases. The complexity and heterogeneity of contemporary urban space poses a challenge to conventional descriptive tools. In recent years, the emergence of urban network analysis and the widespread availability of GIS data has brought network analysis methods into the discussion of urban form. This paper describes a method for computationally identifying clusters within urban and other spatial networks using spectral analysis techniques. While spectral clustering has been employed in some limited urban studies, on large spatialized datasets (particularly in identifying land use from orthoimages), it has not yet been thoroughly studied in relation to the space of the urban network itself. We present the construction of a weighted graph Laplacian matrix representation of the network and the processing of the network by eigen decomposition and subsequent clustering of eigenvalues in 4d-space.In this implementation, the algorithm computes a cross-comparison for different numbers of clusters and recommends the best option based on either the 'elbow method,' or by "eigen gap" criteria. The results of the clustering operation are immediately visualized on the original map and can also be validated numerically according to a selection of cluster metrics. Cohesion and separation values are calculated simultaneously for all nodes. After presenting these, the paper also expands on the 'silhouette' value, which is a composite measure that seems especially suited to urban network clustering.This research is undertaken with the aim of informing the design process and so the visualization of results within the active 3d model is essential. Within the paper, we illustrate the process as applied to formal grids and also historic, vernacular urban fabric; first on small, extract urban fragments and then over an entire city networks to indicate the scalability.
keywords	Urban morphology; network analysis; spectral clustering; computation
series	CAADRIA
email
last changed	2022/06/07 07:59

_id	caadria2020_091
id	caadria2020_091
authors	Ren, Yue and Zheng, Hao
year	2020
title	The Spire of AI - Voxel-based 3D Neural Style Transfer
source	D. Holzer, W. Nakapan, A. Globa, I. Koh (eds.), RE: Anthropocene, Design in the Age of Humans - Proceedings of the 25th CAADRIA Conference - Volume 2, Chulalongkorn University, Bangkok, Thailand, 5-6 August 2020, pp. 619-628
doi	https://doi.org/10.52842/conf.caadria.2020.2.619
summary	In the architecture field, humans have mastered various skills for creating unique spatial experiences with unknown interplays between known contents and styles. Meanwhile, machine learning, as a popular tool for mapping different input factors and generating unpredictable outputs, links the similarity of the machine intelligence with the typical form-finding process. Style Transfer, therefore, is widely used in 2D visuals for mixing styles while inspiring the architecture field with new form-finding possibilities. Researchers have applied the algorithm in generating 2D renderings of buildings, limiting the results in 2D pixels rather than real full volume forms. Therefore, this paper aims to develop a voxel-based form generation methodology to extend the 3D architectural application of Style Transfer. Briefly, through cutting the original 3D model into multiple plans and apply them to the 2D style image, the stylized 2D results generated by Style Transfer are then abstracted and filtered as groups of pixel points in space. By adjusting the feature parameters with user customization and replacing pixel points with basic voxelization units, designers can easily recreate the original 3D geometries into different design styles, which proposes an intelligent way of finding new and inspiring 3D forms.
keywords	Form Finding; Machine Learning; Artificial Intelligence; Style Transfer
series	CAADRIA
email
last changed	2022/06/07 07:56

_id	caadria2020_234
id	caadria2020_234
authors	Zhang, Hang and Blasetti, Ezio
year	2020
title	3D Architectural Form Style Transfer through Machine Learning
source	D. Holzer, W. Nakapan, A. Globa, I. Koh (eds.), RE: Anthropocene, Design in the Age of Humans - Proceedings of the 25th CAADRIA Conference - Volume 2, Chulalongkorn University, Bangkok, Thailand, 5-6 August 2020, pp. 659-668
doi	https://doi.org/10.52842/conf.caadria.2020.2.659
summary	In recent years, a tremendous amount of progress is being made in the field of machine learning, but it is still very hard to directly apply 3D Machine Learning on the architectural design due to the practical constraints on model resolution and training time. Based on the past several years' development of GAN (Generative Adversarial Network), also the method of spatial sequence rules, the authors mainly introduces 3D architectural form style transfer on 2 levels of scale (overall and detailed) through multiple methods of machine learning algorithms which are trained with 2 types of 2D training data set (serial stack and multi-view) at a relatively decent resolution. By exploring how styles interact and influence the original content in neural networks on the 2D level, it is possible for designers to manually control the expected output of 2D images, result in creating the new style 3D architectural model with a clear designing approach.
keywords	3D; Form Finding; Style Transfer; Machine Learning; Architectural Design
series	CAADRIA
email
last changed	2022/06/07 07:57

_id	ecaade2020_167
id	ecaade2020_167
authors	Newton, David, Piatkowski, Dan, Marshall, Wesley and Tendle, Atharva
year	2020
title	Deep Learning Methods for Urban Analysis and Health Estimation of Obesity
source	Werner, L and Koering, D (eds.), Anthropologic: Architecture and Fabrication in the cognitive age - Proceedings of the 38th eCAADe Conference - Volume 1, TU Berlin, Berlin, Germany, 16-18 September 2020, pp. 297-304
doi	https://doi.org/10.52842/conf.ecaade.2020.1.297
summary	In the 20th and 21st centuries, urban populations have increased dramatically with a whole host of impacts to human health that remain unknown. Research has shown significant correlations between design features in the built environment and human health, but this research has remained limited. A better understanding of this relationship could allow urban planners and architects to design healthier cities and buildings for an increasingly urbanized population. This research addresses this problem by using discriminative deep learning in combination with satellite imagery of census tracts to estimate rates of obesity. Data from the California Health Interview Survey is used to train a Convolutional Neural Network that uses satellite imagery of selected census tracts to estimate rates of obesity. This research contributes knowledge on methods for applying deep learning to urban health estimation, as well as, methods for identifying correlations between urban morphology and human health.
keywords	Deep Learning; Artificial Intelligence; Urban Planning; Health; Remote Sensing
series	eCAADe
email
last changed	2022/06/07 07:58

_id	ecaade2020_093
id	ecaade2020_093
authors	Veloso, Pedro and Krishnamurti, Ramesh
year	2020
title	An Academy of Spatial Agents - Generating spatial configurations with deep reinforcement learning
source	Werner, L and Koering, D (eds.), Anthropologic: Architecture and Fabrication in the cognitive age - Proceedings of the 38th eCAADe Conference - Volume 2, TU Berlin, Berlin, Germany, 16-18 September 2020, pp. 191-200
doi	https://doi.org/10.52842/conf.ecaade.2020.2.191
summary	Agent-based models rely on decentralized decision making instantiated in the interactions between agents and the environment. In the context of generative design, agent-based models can enable decentralized geometric modelling, provide partial information about the generative process, and enable fine-grained interaction. However, the existing agent-based models originate from non-architectural problems and it is not straight-forward to adapt them for spatial design. To address this, we introduce a method to create custom spatial agents that can satisfy architectural requirements and support fine-grained interaction using multi-agent deep reinforcement learning (MADRL). We focus on a proof of concept where agents control spatial partitions and interact in an environment (represented as a grid) to satisfy custom goals (shape, area, adjacency, etc.). This approach uses double deep Q-network (DDQN) combined with a dynamic convolutional neural-network (DCNN). We report an experiment where trained agents generalize their knowledge to different settings, consistently explore good spatial configurations, and quickly recover from perturbations in the action selection.
keywords	space planning; agent-based model; interactive generative systems; artificial intelligence; multi-agent deep reinforcement learning
series	eCAADe
email
last changed	2022/06/07 07:58

_id	acadia20_228
id	acadia20_228
authors	Alawadhi, Mohammad; Yan, Wei
year	2020
title	BIM Hyperreality
source	ACADIA 2020: Distributed Proximities / Volume I: Technical Papers [Proceedings of the 40th Annual Conference of the Association of Computer Aided Design in Architecture (ACADIA) ISBN 978-0-578-95213-0]. Online and Global. 24-30 October 2020. edited by B. Slocum, V. Ago, S. Doyle, A. Marcus, M. Yablonina, and M. del Campo. 228-236.
doi	https://doi.org/10.52842/conf.acadia.2020.1.228
summary	Deep learning is expected to offer new opportunities and a new paradigm for the field of architecture. One such opportunity is teaching neural networks to visually understand architectural elements from the built environment. However, the availability of large training datasets is one of the biggest limitations of neural networks. Also, the vast majority of training data for visual recognition tasks is annotated by humans. In order to resolve this bottleneck, we present a concept of a hybrid system—using both building information modeling (BIM) and hyperrealistic (photorealistic) rendering—to synthesize datasets for training a neural network for building object recognition in photos. For generating our training dataset, BIMrAI, we used an existing BIM model and a corresponding photorealistically rendered model of the same building. We created methods for using renderings to train a deep learning model, trained a generative adversarial network (GAN) model using these methods, and tested the output model on real-world photos. For the specific case study presented in this paper, our results show that a neural network trained with synthetic data (i.e., photorealistic renderings and BIM-based semantic labels) can be used to identify building objects from photos without using photos in the training data. Future work can enhance the presented methods using available BIM models and renderings for more generalized mapping and description of photographed built environments.
series	ACADIA
type	paper
email
last changed	2023/10/22 12:06

_id	ecaade2020_017
id	ecaade2020_017
authors	Chan, Yick Hin Edwin and Spaeth, A. Benjamin
year	2020
title	Architectural Visualisation with Conditional Generative Adversarial Networks (cGAN). - What machines read in architectural sketches.
source	Werner, L and Koering, D (eds.), Anthropologic: Architecture and Fabrication in the cognitive age - Proceedings of the 38th eCAADe Conference - Volume 2, TU Berlin, Berlin, Germany, 16-18 September 2020, pp. 299-308
doi	https://doi.org/10.52842/conf.ecaade.2020.2.299
summary	As a form of visual reasoning, sketching is a human cognitive activity instrumental to architectural design. In the process of sketching, abstract sketches invoke new mental imageries and subsequently lead to new sketches. This iterative transformation is repeated until the final design emerges. Artificial Intelligence and Deep Neural Networks have been developed to imitate human cognitive processes. Amongst these networks, the Conditional Generative Adversarial Network (cGAN) has been developed for image-to-image translation and is able to generate realistic images from abstract sketches. To mimic the cyclic process of abstracting and imaging in architectural concept design, a Cyclic-cGAN that consists of two cGANs is proposed in this paper. The first cGAN transforms sketches to images, while the second from images to sketches. The training of the Cyclic-cGAN is presented and its performance illustrated by using two sketches from well-known architects, and two from architecture students. The results show that the proposed Cyclic-cGAN can emulate architects' mode of visual reasoning through sketching. This novel approach of utilising deep neural networks may open the door for further development of Artificial Intelligence in assisting architects in conceptual design.
keywords	visual cognition; design computation; machine learning; artificial intelligence
series	eCAADe
email
last changed	2022/06/07 07:55

_id	acadia20_272
id	acadia20_272
authors	del Campo, Matias; Carlson, Alexandra; Manninger, Sandra
year	2020
title	How Machines Learn to Plan
source	ACADIA 2020: Distributed Proximities / Volume I: Technical Papers [Proceedings of the 40th Annual Conference of the Association of Computer Aided Design in Architecture (ACADIA) ISBN 978-0-578-95213-0]. Online and Global. 24-30 October 2020. edited by B. Slocum, V. Ago, S. Doyle, A. Marcus, M. Yablonina, and M. del Campo. 272-281.
doi	https://doi.org/10.52842/conf.acadia.2020.1.272
summary	This paper strives to interrogate the abilities of machine vision techniques based on a family of deep neural networks, called generative adversarial neural networks (GANs), to devise alternative planning solutions. The basis for these processes is a large database of existing planning solutions. For the experimental setup of this paper, these plans were divided into two separate learning classes: Modern and Baroque. The proposed algorithmic technique leverages the large amount of structural and symbolic information that is inherent to the design of planning solutions throughout history to generate novel unseen plans. In this area of inquiry, aspects of culture such as creativity, agency, and authorship are discussed, as neural networks can conceive solutions currently alien to designers. These can range from alien morphologies to advanced programmatic solutions. This paper is primarily interested in interrogating the second existing but uncharted territory.
series	ACADIA
type	paper
email
last changed	2023/10/22 12:06

_id	acadia20_218
id	acadia20_218
authors	Rossi, Gabriella; Nicholas, Paul
year	2020
title	Encoded Images
source	ACADIA 2020: Distributed Proximities / Volume I: Technical Papers [Proceedings of the 40th Annual Conference of the Association of Computer Aided Design in Architecture (ACADIA) ISBN 978-0-578-95213-0]. Online and Global. 24-30 October 2020. edited by B. Slocum, V. Ago, S. Doyle, A. Marcus, M. Yablonina, and M. del Campo. 218-227.
doi	https://doi.org/10.52842/conf.acadia.2020.1.218
summary	In this paper, we explore conditional generative adversarial networks (cGANs) as a new way of bridging the gap between design and analysis in contemporary architectural practice. By substituting analytical finite element analysis (FEA) modeling with cGAN predictions during the iterative design phase, we develop novel workflows that support iterative computational design and digital fabrication processes in new ways. This paper reports two case studies of increasing complexity that utilize cGANs for structural analysis. Central to both experiments is the representation of information within the data set the cGAN is trained on. We contribute a prototypical representational technique to encode multiple layers of geometric and performative description into false color images, which we then use to train a Pix2Pix neural network architecture on entirely digital generated data sets as a proxy for the performance of physically fabricated elements. The paper describes the representational workflow and reports the process and results of training and their integration into the design experiments. Last, we identify potentials and limits of this approach within the design processes.
series	ACADIA
type	paper
email
last changed	2023/10/22 12:06

_id	cdrf2019_103
id	cdrf2019_103
authors	Runjia Tian
year	2020
title	Suggestive Site Planning with Conditional GAN and Urban GIS Data
source	Proceedings of the 2020 DigitalFUTURES The 2nd International Conference on Computational Design and Robotic Fabrication (CDRF 2020)
doi	https://doi.org/https://doi.org/10.1007/978-981-33-4400-6_10
summary	In architecture, landscape architecture, and urban design, site planning refers to the organizational process of site layout. A fundamental step for site planning is the design of building layout across the site. This process is hard to automate due to its multi-modal nature: it takes multiple constraints such as street block shape, orientation, program, density, and plantation. The paper proposes a prototypical and extensive framework to generate building footprints as masterplan references for architects, landscape architects, and urban designers by learning from the existing built environment with Artificial Neural Networks. Pix2PixHD Conditional Generative Adversarial Neural Network is used to learn the mapping from a site boundary geometry represented with a pixelized image to that of an image containing building footprint color-coded to various programs. A dataset containing necessary information is collected from open source GIS (Geographic Information System) portals from the city of Boston, wrangled with geospatial analysis libraries in python, trained with the TensorFlow framework. The result is visualized in Rhinoceros and Grasshopper, for generating site plans interactively.
series	cdrf
email
last changed	2022/09/29 07:51

_id	artificial_intellicence2019_117
id	artificial_intellicence2019_117
authors	Stanislas Chaillou
year	2020
title	ArchiGAN: Artificial Intelligence x Architecture
source	Architectural Intelligence Selected Papers from the 1st International Conference on Computational Design and Robotic Fabrication (CDRF 2022)
doi	https://doi.org/https://doi.org/10.1007/978-981-15-6568-7_8
summary	AI will soon massively empower architects in their day-to-day practice. This article provides a proof of concept. The framework used here offers a springboard for discussion, inviting architects to start engaging with AI, and data scientists to consider Architecture as a field of investigation. In this article, we summarize a part of our thesis, submitted at Harvard in May 2019, where Generative Adversarial Neural Networks (or GANs) get leveraged to design floor plans and entire buildings .
series	Architectural Intelligence
email
last changed	2022/09/29 07:28

For more results click below: