Image: Sidewalk Labs

Technology

Machine learning used in generative tool for city planning

5 January 2020

A new generative tool developed by a Toronto-based smart cities specialist uses machine learning and computational design to generate millions of comprehensive planning scenarios.

Sidewalk Labs, the urban innovation offshoot of Google holding company Alphabet Inc, has developed the design tool that applies machine learning and computational design to aid urban planning and architecture. 

Sidewalk Lab’s product manager Violet Whitney and designer Brian Ho explain in a blog that the machine learning produces a planning process that is “more holistic and efficient, helping planners and the community make the most informed decision possible”.

The tool can generate millions of comprehensive planning scenarios considering various data sets, such as zoning regulations, existing space and physical qualities. Sidewalk Labs can then analyse those scenarios for various quality of life criteria, such as density, weather patterns, economics and walkability.

The generative design tool will be tested on Sidewalk Labs’ Quayside project in Toronto (Image: Sidewalk Labs)

The blog’s authors say the generative design tool is not intended to replace designers, but rather give them more planning options the ability to consider more data than is currently possible. For example, the tool could help designers to see how one design option might maximise density but inadvertently block sunlight, or show how a creative design could best use existing space.

The tool, Whitney and Ho write, could also improve opportunities for community engagement and generate more discussion of city plans.

It also fits into Sidewalk Labs’ goal to bring machine learning and flexibility to the urban design process, they say.

This approach will be put to the test in the company’s Quayside project in Toronto. The development showcases not only data collection technology, but also the potential of smart development, such as underground waste collection, weather-adapting buildings and dynamic street pavements that can be modified to fit different needs.

Sidewalk Labs has also worked on street design principles, encouraging developers and planners to adapt to new mobility options by tailoring streets for different modes, separating them based on speed and introducing more flexibility. 

The generative design tool will bring some of these same principles to the urban planning and zoning realm, helping developers make informed decisions about buildings from the start.

Other cities have incorporated data software and 3D mapping to their zoning processes to help developers understand how their plans will impact citizens, and the Sidewalk Labs tool can take that a step further.