AI Tool Developed by AUAS Inspires Architects and Urban Planners

The AI tool traces high-density urban locations, making it easier to find appropriate reference locations

Sunday, 15 August 2021

The Netherlands intends to build one million new dwellings over the next few years. Most of these will be built within the existing urban fabric. How can this densification also be used to create favourable living environments? To answer this, reference locations are necessary to see how design solutions can turn out. They also play an important role in the processes between spatial designers and their clients. As part of the Sensing Streetscapes research project within the Spatial Urban Transformation and Responsible IT research programmes, the Amsterdam University of Applied Sciences (HvA) has used open data to develop an AI tool that automates the search for reference locations.

 

 

Amsterdam, too, is on the cusp of extreme densification. Tens of thousands of new dwellings will be built in the Amsterdam metropolitan region, most of them within the city’s boundaries. This will result in high-density building, often combined with high rise. Each development area’s size, location and programme are unique. Moreover, the new neighbourhoods must meet a growing number of requirements and wishes. The AI tool makes it possible to quickly trace comparable locations. How were the design tasks solved there? Which design solutions were applied, what type of programme was used, and what was the extent of densification? And has this succeeded in creating an attractive and functional environment? An open-access search machine enables architects, urban planners and developers to filter so they can search according to a living environment’s conditions for which they require information.

The AI tool explores high-density, high-rise environments with a mixed-use profile (living, working and leisure, FSI, GSI etc). During realisation, however, this was easier said than done. ‘Data hasn’t yet changed the discipline of spatial design as much as other sectors. This was the very reason why we also wanted to include this line of experiment in our research,’ says Professor of Spatial Urban Transformation, Frank Suurenbroek. ‘And it has been a success,’ adds the project’s leader, Gideon Spanjar, ‘even though the tool only provides a basis for a more refined search.’

The Development of the AI Tool

First, the Copernicus Urban Center’s dataset set was used to divide the world into a 250×250 metre grid and to add basic characteristics to the grid. Next, the OpenStreetMap database was loaded onto our servers and linked to the grid. OpenStreetMap is an open-source platform where volunteers add and annotate geographic details. In addition, the AI tool was ‘fed’ with suitable international examples from the world of architecture, partly obtained through an appeal on ArchDaily, the leading architects’ blog.

‘Based on the map information and the features of the international examples, relevant indicators were formulated that tell something about the neighbourhood, block or buildings,’ says Maarten Groen, who built the AI tool and who is a senior researcher for the Responsible IT research programme at the university’s Centre of Applied Research Technology Create-IT. ‘Think of a building’s height, the height to width ratio, how much greenery and water are present in the neighbourhood, as well as the number of shops, restaurants and public transport facilities: these are all variables that can tell us something about the quality of the living environment in urban locations.’ Using all this input, the tool’s algorithm can learn to recognise patterns in the data (machine learning) so that it understands what is ‘right’ when it sees comparable urban developments. Maarten Groen explains: ‘With this information, we can “unleash” the tool to also find less well-known places of interest. This is how we have automated the process for finding reference locations right across Europe and North America.’

More Information

The Sensing Streetscapes AI tool is a collaboration between Frank Suurenbroek, Professor of Spatial Urban Transformation at the Centre of Expertise Urban Technology, and Nanda Piersma, Professor of Responsible IT at the Centre of Applied Research Technology Create-IT.