Generative design for architectural spatial layouts: a review of technical approaches

The design of spatial layouts is a classic challenge in architectural design. This study reviews the current research on generative design for architectural spatial layouts, focusing mainly on technological methodologies. These methodologies can be broadly categorized into two main approaches: data-...

Full description

Saved in:
Bibliographic Details
Main Authors: Shurui Yan, Chen Wu, Yixin Zhang
Format: Article
Language:English
Published: Taylor & Francis Group 2025-06-01
Series:Journal of Asian Architecture and Building Engineering
Subjects:
Online Access:http://dx.doi.org/10.1080/13467581.2025.2512235
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The design of spatial layouts is a classic challenge in architectural design. This study reviews the current research on generative design for architectural spatial layouts, focusing mainly on technological methodologies. These methodologies can be broadly categorized into two main approaches: data-driven and knowledge-driven. Initially, the study defines each approach by elucidating their principles, introducing typical methods, providing relevant research statistics, contrasting the characteristics of various methodologies, and critically analysing the advancements and constraints inherent in each technical approach. In conclusion, for the data-driven approach, key research areas include expanding datasets, enhancing alignment with human values, and improving model explainability. For the knowledge-driven approach, finding effective solutions to translate architectural design knowledge into programs is essential. Additionally, improving computational efficiency makes knowledge-driven methods, such as heuristic algorithms, more user-friendly. Furthermore, the study suggests that an intelligent combination of these two approaches at different phases of generative design yields a promising research direction. This synergy compensates for their respective limitations and leverages the full potential of human expertise encapsulated within data and rules, thereby addressing insufficient data and ensuring that models align well with human values.
ISSN:1347-2852