A group of The New Normal researchers prototype a deep learning tool for smart cities based on a simulation of the urban environment.
The urban design and governance platform Atoll is a deep learning model that simulates an interplay between complex urban systems. It is a tool for urban data aggregation, management, and interpretation developed by Tom Pearson, Leo Stuckardt, Artem Konevskikh, and Natalie Mezhetskaya as part of Strelka Institute’s The New Normal Program 2018.
Atoll is able to simulate and thus predict human flows and activities in certain urban environments. It is an open-source project, aimed towards a more inclusive and participatory urbanism. The platform is similar to already existing Building and City simulation models (BIM and CIM). The difference is that Atoll is intended to create protocols for engagement with multiple urban simulations.
“As we move through cities, utilize digital services, and are being surveyed by CCTV and sensor networks, we are becoming an implicit part of multiple models of the city – corporate models and public ones alike. Yet the design and intentions behind most of those models is unclear,” explain the researchers in their statement. “Atoll offers not only transparency of when and where your data is collected but also the opportunity to shape the city through your own urban models. Since the platform is open source, we will never share or monetize your personalized data.”
At the core of Atoll lies a data management engine – Atoll MIM. It stores two sets of data – GIS (Geospatial Information System) and behavioral data. GIS data includes all properties of a physical location, including roads, buildings, parks. Behavioral data collects dynamic parameters like movements, taxation zones, and opening hours. All of this data is then anonymized through machine learning-driven compression algorithms.
“We believe that cities are reaching new levels of complexity as urban planning processes and regulations, traditionally held in public hand, are being increasingly influenced by platforms and digital services. In addition, sensor networks and digital information systems allow us to read the city in more detail. Since these enormous, fluctuating amounts of data exceed human comprehension, Atoll aims to employ machines to distinguish and classify data constellations," the researchers state.
Atoll is currently in the alpha development stage. You can follow updates on the project’s website.