Traditionally one city is planned and designed by static master planning regulation. Paradigm established from last century (even updated every 5-10 years) becomes increasingly harder to propose dynamic strategy and evaluation in this rapid urban development. To provide efficient and direct reference of a well-planned city for a new site becomes the issue that needs to be answered. Therefore, an intelligent urban design system is urgently needed in city optimization.
This course, That was part of Digital FUTURES Shanghai Summer Workshop, designs a set of Al urban design system for one existing city/imagined site, focuses on building participants’ basic skill in artificial intelligence model, and the application of urban spatial data. we explored how to utilize the existing data type, learn to extract feature and annotations of targeted city, set these parameters in provided machine learning and deep learning model, thereby run the model and obtain experimental results. These results will be graded by a prior evaluation system.
The main purpose of the AI Urban Design was to analyze and understand the way in which people perceive sets of public urban spaces through the form of visual data. Due to rapid urban development, it becomes increasingly harder to provide an urban layout that is functional for all. Therefore, it is crucial that urban designers know how to adjust public spaces into an efficient and productive manner in order to maintain these continuously developing cities.
This will be done through a binary classifier using machine, which analyzes human perceptions of spaces. For example, the design of a fountain in the center of Dizingoff Square in Tel Aviv has been a controversial topic to the public, and as a result, the continuation of the fountain has been up for debate. The comparison between the fountain and a building in this space has been a prevalent issue among urban designers. AI wants to use this binary classifier machine in order to better understand what percentage of people like the fountain vs the building, as well as the impact each structure has to the surrounding lot. This information will help them to determine the proper approach to take on any potential renovations of this public urban area.
The design of the fountain in the center of the space has been controversial and its future is subject to debate in future renovation of the public space.
Can we understand by deep learning image analysis
How people perceive the space?
And how significant the fountain is to public perception of the space?
Visual data sets capture latent information which suggests to human perceptions of public space.
Deep learning offers a new way to mine information embedded in images of the space from the visual data sets of Social networks.
Employing CNNs to harness data offers a new technique for public feedback
and crowd sourcing – or cloud sourcing -- architectural and urban planning decisions with latent feedback held in images posted in the digital commons via Instagram, Facebook, Twitter etc. This exploration offers novel possibilities to urban planners and public policy decision makers as they seek to find ways to enhance public space.
Group Members Aaron Berke, Jared Grogan , Hanan Peretz
**Instructors**
1. He WanYu
Founder and CEO, Xkool. Tech, China
2. Jackie Yong Leomg Shong
Senior Researcher, Future Architect Lab, China