Sidewalk Labs Brings Machine Learning to Urban Planning

Machine learning enables Delve to produce high-performing design variants that take into account a multitude of factors, such as usable built area, daylight access, exceeded all goals across unit yield, rental pricing recommendations and much more. (Image courtesy of Sidewalk Labs.)
The urban population is increasing across the globe. More than 55 percent of the world’s population lives in cities today and 68 percent are expected to do so by 2050. That’s a challenge for urban developers. Incorporating sustainability with quality-of-life concerns while maximizing land space and infrastructure is a noble goal for the conscientious urban planner and Sidewalk Labs developed Delve to help with that. Since it was launched in 2020, Delve may have become the type of generative design tool needed to digitally create the cities of the future.

Urban Development Planning Challenges

Planning a city isn’t as simple as popular video games make it out to be. The development team includes a host of specialists, from architects to developers and planners, who must look at every aspect of what a neighborhood should be, both now and in the future. They must consider the space available, increased traffic, buildings casting shadows over public spaces, future housing needs and the impact on people, just to name a few—all while creating a neighborhood that must last for decades.

As these teams set out to create something new, they often employ different software applications to design, plan streets and maintain diligent financial records. Mindful of the effort and inefficiencies involved in using multiple applications, Sidewalk Labs developed Delve to provide a holistic approach that includes multiple design options and takes into account the many constraints developers face.

Delve: An Integrated Generative Design Solution

Along with time and monetary constraints, a development project must consider the whole long life of the project. Traditional tools and methods were not up to the task, according to Sidewalk Labs. The company designed Delve for creating the best possible neighborhood design based on set priorities for a specific project. According to Sidewalk Labs, “Delve can quickly identify thousands of design options and priority outcomes, such as sunlight hours, walkability and views—even down to the unit level—then optimize for the best possible design and financial performance.”

Crafted with integration in mind, Delve is suitable to use at any or all stages of development. It brings together the different teams into one platform, which increases the chances of better decision-making, provides the ability to make quick changes, and saves time by not having to rely on repetitive modeling.

At the start of a project, the discovery process involves inputting plan milestones and constraints as well as prioritizing outcomes. Delve works by translating those complex requirements into an optimization problem. Typical problem sets that are identified and considered include:

  • Inputs: Physical design and development program strategies that will be varied or explored
  • Constraints: Requirements required for acceptance
  • Priority outcomes: Key performance metrics that indicate success
  • Benchmarks: Baseline designs for basis comparison

With that data, Delve will literally create thousands of potential designs while providing needed impact measurements. It will sort them based on the following criteria and the development team can sift through the top designs to present. Instead of taking months or longer to produce design options, the development team can have them in minutes.

Features of Delve include:

  • Ranked Options: High-performing design options are ranked according to priority metrics performance. Delve produces a report that makes it easier for the development team to navigate the many design options.
  • High-fidelity Designs: The generation of these designs can be allocated to individual sites to discover additional residential density, as well as provide evaluations for walkability, view, daylight access and more for each unit.
  • Integrated Financial Models: From the open spaces to the buildings and street sections, there is a cost associated with each. Delve’s financial model keeps track of the individual segments, providing cost models that can be seamlessly integrated into a spreadsheet.
  • Built-in Utility Models: Delve makes it easier to meet sustainability goals by providing ongoing energy measurements, including water use, waste and solar intake.

Use Case Studies

Does Delve make a difference? After it was first released, a Japanese real estate development team worked with Sidewalk Labs to determine the best scenarios for a large-scale development in a major city. The site was being designed to include parks, offices and hotels. With its AI power, Delve created more than 10,000 scenarios with prioritized parameters that included public space, height limits, additional local transit and an adjacent parcel. Each scenario included an evaluation of profitability, enabling more informed real-time decisions. By using the integrated financial model with machine learning, key life metrics, such as views and walkability, were included to narrow down the options.

A prominent Japanese real estate development team used Delve to test different design scenarios for parcels, public space and height limits while evaluating financial return and quality of life. (Image courtesy of Sidewalk Labs.)
According to Delve, after two months, 50 percent of the generated designs exceeded the benchmarks for the development team’s financial goals and quality-of-life outcomes. Delve also created scenarios for an adjacent development parcel.

Once the developers provided their inputs, Delve recommended a reduction of the office and hotel by 10 percent and 12 percent and creating 40,000 square meters of residential space. The hotel and office had improved views of the parks and water by more than 40 percent. Delve also produced evidence for raising the rent 5 percent higher than originally planned.

Refining price points from Delve recommendations resulted in the developer underwriting a rent premium of 9.8 percent, which was an improvement of 370 basis points. Delve enabled the weighing of these options in real time, which would have likely taken months to achieve.

A rendering of the 85-acre master plan for the Wembley Park neighborhood. (Image courtesy of Quintain.)
Based in the United Kingdom, developer Quintain looked to Delve to start planning an innovative 12-acre mixed-used development in Wembley Park. The thousands of built-to rent and multifamily homes were to consider quality-of-life metrics, which Quintain had been unable to do between 2016 and 2019.

Quintain aimed to increase housing density without sacrificing open spaces or blocking the daylight. The constraints were meeting height restrictions and ensuring 3.5 acres of open space with priority outcomes of building daylight access, residential unit yield, ground sunlight access, daylight access for neighboring buildings, construction costs and value.

After inputting the priorities and constraints, Delve initially created, optimized and evaluated more than 40,000 variants. From there, 24 high-performing design options were created. All of them met the quality-of-life benchmarks. In one of the included designs, the developer was able to increase the number of units built by 200 while also making them larger by 13 square feet with enhanced daylight access. This variant resulted in an 11 percent increase in open space.

Thanks to Delve, the project was able to exceed benchmark performance. The generative tool identified optimal design strategies that included developing oddly shaped areas of the land into courtyard blocks and purposefully arranging blocks for increased daylight access. Building heights were able to be increased in certain locations without impacting neighboring development.

As the urban sprawl continues, it creates challenges across the board for both planning urban areas, their construction and addressing infrastructure issues, such as water. Digital tools like Delve are offering new ways to integrate the planning and design process, creating a new path for designing the future that considers sustainability and quality of life.