Trajectory of building and structural design automation from generative design towards the integration of deep generative models and optimization: A review

Abstract

Design automation addresses the problem of manual and time-consuming design processes by streamlining and optimizing the creation of products, systems, or solutions using automated tools and workflows. The insufficient implementation of design automation technologies in the building and structural design industry has significant implications for productivity, innovation, and sustainability. This underscores an urgent need for comprehensive solutions to address this societal challenge. In this paper, a review of building and structural design automation methods is presented, covering the progression from generative design to the integration of deep generative models and optimization techniques. This review explores generative design and deep generative models, including generative adversarial network, variational autoencoder, and reinforcement learning frameworks, followed by topology optimization and their applications. This paper discusses key works that have introduced methods and techniques in generative design. Moreover, it addresses key challenges observed in generative systems across design fields. Furthermore, potential opportunities for future research are also identified, highlighting crucial areas that require strategic solutions. This study aims to establish a comprehensive reference framework to facilitate method selection and enhance existing methodologies through a systematic hybridization approach, thereby advancing the capabilities of the field.