Considerations when Preparing for and Deploying to New Cities

Project Sidewalk is an open source project run by Professor Jon E. Froehlich, research scientist Michael "Mikey" Saugstad, and PhD and undergraduate students at the University of Washington.

From the beginning, our overarching vision has been to develop tools & techniques that semi-automatically map and assess all of the sidewalks in the world—using an intermixing of computer vision, machine learning, and crowdsourcing. As part of this vision, we are looking for new deployment cities outside of the US.

Based on our experiences deploying into cities in the US, the Netherlands, Mexico, Zurich, and beyond, here are our primary considerations:

• Strong, active partners. Most importantly, the presence of strong local advocates such as citizens, non-profit organizations, or government officials who can help us recruit and sustain communities of volunteers to help (e.g., to help run mapathons, to engage local schools and youth organizations, etc.). We want to partner with you to make this work. Ideally, we would also work with people with mobility disabilities in our deployment cities to help inform how to deploy and improve our tools and help impact change. As some examples:

  • In Newberg, Oregon we worked directly with local community accessibility advocates and the city government
  • In San Pedro Mexico, we worked with a Mexico-based NGO called Liga Peatonal and the local government
  • In Amsterdam, we worked with the local government and the organization World Enabled
  • In Oradell, New Jersey we worked with local Girl Scout troops, the local MS society, and the local hospital

• City size. smaller cities tend to have fewer resources and thus may benefit more from Project Sidewalk. If your city is large, we may want to do a slow-roll deployment where we start with a smaller subsection (e.g., 100-120 miles or 160-193 kilometers). Typically, our users can label at a rate of ~1.8 mph though this is based on expertise and skill.

• Need. Are the sidewalks in your city currently in https://www.openstreetmap.org/? Does your city employ on-the-ground workers for physical accessibility inspection? Is this information publicly available (e.g., on an open government website)? What is the state of your pedestrian infrastructure?

• The presence of and recent imagery from Google Street View (GSV). Project Sidewalk uses Google Street View for imagery. Thus, our techniques work best when a GSV car has recently driven through your city. If you're not sure about this, contact us and we can investigate for you!

• Native language. Project Sidewalk was designed in English but our partners have helped us natively translate our interface into Spanish (Mexico), Dutch, German, and Chinese. If you would like to deploy Project Sidewalk in your city with a different language, we will need your help with language translations.

• Ideally, financial support. This is a discussion we should have over email or video conferencing but ideally there would be some tangible resources (either via personnel time or money) that you could supply

In addition, because Project Sidewalk relies on neighborhood geographies for routing missions, we will need neighborhood boundary data in ArcGIS shapefile format or GeoJSON. See this conversation about Teaneck, NJ neighborhoods on GitHub (link).

Because deploying into a new city is not (yet) completely automated, we estimate that it takes @misaugstad (our lead dev and research scientist) about ~2-4 days per city to setup and configure the city.