BBS, or the Brown Bag Seminar, is an academic exchange initiative launched by the Tsinghua Urban Ergonomics Lab with support from many research groups. It serves young scholars committed to research and interested in developments at the frontiers of their fields. Sessions are held from time to time at the Urban Ergonomics Lab in the north wing of the Architecture Building at Tsinghua University.

How can generative AI help urban designers create future urban landscapes?

As generative artificial intelligence develops, its potential applications in urban design are becoming increasingly apparent. However, existing automated approaches often lack mechanisms for involving human experts, making it difficult to generate urban spaces that are both feasible in practice and functionally coherent.

In this session of Academic BBS, Yuebing Liang, assistant professor at the School of Architecture, Tsinghua University, will give a presentation entitled Urban Landscape Generation Through Human-AI Collaboration: Exploring Multimodal Diffusion Models. The research proposes a framework for generating built environments through human-AI collaboration. Based on modified Stable Diffusion and ControlNet models, it generates multimodal urban elements in stages, including land-use layouts, building-form configurations, and satellite imagery. At each stage, the AI model accepts input from human experts, enabling an interactive process guided by people and generated by machines.

The study shows that this approach outperforms conventional GAN models in image quality, alignment with human input, and design diversity, making generative urban design more useful and easier to apply in practice. The research demonstrates the advantages of diffusion models for human-AI collaborative urban design and offers a viable path and a new paradigm for developing future intelligent urban design tools.□

About the speaker

Yuebing Liang is an assistant professor and doctoral supervisor in the Department of Urban Planning and Design, School of Architecture, Tsinghua University. She previously worked as a postdoctoral associate at the Singapore-MIT Alliance for Research and Technology. She holds a PhD in urban planning from The University of Hong Kong and bachelor's and master's degrees in architecture from Tsinghua University. Her research lies at the intersection of artificial intelligence and transport and urban planning, encompassing spatiotemporal intelligence and human mobility modeling, applications of generative AI and reinforcement learning in urban and transport planning, and big-data analysis of travel behavior.