Clear the Way

Rick Eldridge knows to choose an alternate path around the massive tree root buckling the sidewalk near his Northwest Washington, D.C., home. But when he ventures in his electric wheelchair beyond familiar streets, he can’t predict what impediments he might encounter.

That’s why Eldridge, 48, who was paralyzed decades ago when his car went off a rain-slicked highway, is helping University of Maryland computer science researchers test Project Sidewalk. The initial goal is to create crowd-sourced map of D.C. sidewalk accessibility—and then to automate the process to eventually scope out the sidewalks all over the world and guide people with mobility limitations along safe routes.

Anyone can help by signing up at the project website and simply viewing Google Street View images and digitally highlighting obstacles like blocked sidewalks or non-existent curb ramps.

The website provides a tutorial to get started and is meant to be game-like despite its serious purpose, says Jon Froehlich (left), an assistant professor of computer science with a joint appointment in the UMD Institute for Advanced Computer Studies and founder of Project Sidewalk.

“Give me 15 minutes of your time, and you can audit a mile of street for us,” he says.

Since the website went live in the fall, about 430 registered users have virtually inspected the accessibility of 450 miles of roadway in D.C., meaning the project is just over 40 percent of the way to its mileage goal.

The project was conceived to meet the needs of people like Eldridge, whom Froehlich met while attending the University of Washington-Seattle and reconnected with after both moved to D.C., but it can help everyone, Froehlich says.

“Traffic data is copious, but there’s not much to help you move around the city if you’re not in a car,” Froehlich says. “We’ve all pulled luggage or pushed a stroller, and if there’s no sidewalk, it affects my ability to move around the world outside.”

For people with disabilities that limit their mobility and weaken their health, inaccessible sidewalks are more than a source of anxiety, says Eldridge. “That you could be late somewhere isn’t the problem,” he says. “Imagine you come up on a barrier you can’t get past and you have to retrace your pathway and find another route. And then imagine it’s 95 degrees and humid, or it’s freezing or there’s a significant rainstorm. It can expose you to a serious health situation.”

Low batteries on his wheelchair, particularly late in the day, add another layer of complexity for getting around, he says.

One of the goals of the project is to develop “smart routing,” says Manaswi Saha, a second-year Ph.D. student. This would allow users to follow accessible routes just as drivers use Google Maps or Waze for turn-by-turn directions.

The system would also provide an “access score”—a number to indicate the general sidewalk accessibility of a neighborhood or city.

Researchers are still depending on participants to help them fill in the blank spots on the map in D.C., but not forever. The users auditing D.C. streets are additionally helping develop an automated system that will scour Street View data or other imagery to generate maps that will only require people to spot-check for accuracy. The automatic audits will be far speedier than manually inputting new data.

“Every time you click on the website, you’re training a machine through a process called machine learning,” a type of artificial intelligence that involves computers able to learn and integrate new data without additional programming, Froehlich says. “For speed, nothing can beat an algorithm.”

Froehlich likens Project Sidewalk to a tech startup firm, rife with technical challenges and overflowing with potential applications.

“There are so many possibilities, I could probably work on this for another 10 years,” he says.