Project PI : Immanuel Koh

Project Period : 01 December 2020 to 30 November 2022

The ability to infer and generate unknown interior spatial configurations from any perceptible exterior building features and urban/rural contexts poses great opportunities in harnessing artificial spatial intelligence. The objective of the project is the AI generation of indoor floor plans directly from building exteriors based on the underlying design correlations between a building’s external form and its internal spaces. Both photogrammetry and deep neural networks will be used to develop the proposed system.

Potential applications (both commercial and military)

Other non-defence use cases may include intelligent building façade recognition for navigating drones and autonomous vehicles, automated building defects detection for building maintenance and rapid generative design for real estate development.

Roadmap of the technology

The project will progressively include more diverse (and complex) building and city forms. This tool will be developed and made transferrable to other related spatial domains.

Project team’s current expertise and relevant research achievements

This research is supported and informed by expertise drawn from the domains of artificial intelligence, computer vision, remote sensing, , generative design, architectural and urban planning.