8 Aug

Project Sidewalk Spotlight

At A Glance

Project Sidewalk is a research group working in the Makeability Lab of the University of Maryland with the aim of mapping accessibility in cities to aid those with mobility-impairments. So far, they have used crowd-sourced data gathering techniques to map 51% of the D.C. Metro area and their goal is to have 100% of the area mapped before branching out to other cities across the United States. By mapping accessibility, the Project Sidewalk team looks to aid in the creation of better routing and city planning to ensure that freedom of mobility is available to all, regardless of their impairments.

In Depth

Have you ever walked through your neighborhood and seen a fire hydrant awkwardly placed in the middle of the sidewalk? Or maybe noticed that you needed to step up onto an unusually high curb after crossing a street. For some of us, these obstacles are only minor inconveniences. However, for more than 30 million physically impaired Americans, these obstacles are much more than just a minor annoyance.

Luckily, the Project Sidewalk team has been working hard on solutions. Project Sidewalk is a web application that uses online auditing and machine learning to map accessibility. The group is led by Jon Froehlich, a Professor of Computer Science at the University of Maryland. Currently, his team consists of 3 graduate students, 4 undergraduate students, and 1 high school student. With just this small team, they have managed to map half of the D.C. Metro area and are hopeful that they will have 100% of D.C. mapped by the end of the year.

According to Froehlich, mapping accessibility is something that he is both professionally and personally invested in. As a computer scientist, he has been fascinated by the immense amount of data made publicly available through Google Street View and had been looking for a way to do something meaningful with it. As a father of two, he understands that accessibility can be a big issue for parents pushing young children in strollers. With the Project Sidewalk application, he has found a way to combine these needs within a tool that helps not only with toddler transportation, but also with the mobility-impaired.

Before the Project Sidewalk tool existed, collection of accessibility information was done by in-person auditing, which was laborious and time-consuming. Instead, the Project Sidewalk tool takes advantage of Google Street View’s street images to have users on the internet point out accessibility. Accessibility is not only categorized by type (such as curb cuts, ramps, etc.) but also by the quality of the accessibility (on a scale of 1 to 5).

The Project Sidewalk auditing system. Click on the image to try it out


The website introduces the user to the tool with an easy-to-navigate tutorial that gives step-by-step instructions on how to properly audit an intersection.

According to a report released by the Project Sidewalk team, 185 crowd workers were able to correctly identify the presence of various accessibility problems with 81% accuracy when an individual audited an area and up to 90% accuracy when 5 or more people audited the same area. This data provided sufficient evidence to support the claim that Google Street View based auditing is a viable method of mapping accessibility.

So what’s in store for the future of the Project Sidewalk team? After completing 100% of the D.C. area, they are looking to improve upon their machine learning system, ‘Tohme’, with the goal of making Project Sidewalk a large-scale, efficient, and accurate tool for auditing streets in every city. As of now, the machine learning system is still young but has already reduced human-time-cost in mapping accessibility by 13%. With more complex algorithms and a larger data pool of ground truth data, they hope to reduce this time cost by one order of magnitude.

We encourage all of our readers to try out the Project Sidewalk software and aid them in mapping accessibility for all!

Written by Lewis Berger