Where the sidewalk ends – Zoo House News

It’s easier than ever to view maps of any place you want to visit – that is, by car. Walking is another matter. Most cities in the US don’t have street maps, and pedestrians are usually left to their own devices: Can you walk from your hotel to the restaurants across the freeway? Is there a shortcut from downtown to the sports arena? And how do you even get to this bus stop?

Now, MIT researchers, along with colleagues from several other universities, have developed an open-source tool that uses aerial photography and image recognition to create full maps of sidewalks and crosswalks. The tool can help planners, policy makers and urban planners looking to expand pedestrian infrastructure.

“In the fields of urban planning and city policy, this is a huge gap,” says Andres Sevtsuk, associate professor at MIT and co-author of a new paper detailing the tool’s capabilities. “Most US municipalities know very little about their sidewalk networks. There is no data about it. The private sector has not taken on the task of mapping them. It seemed to be developing a really important technology, especially in an open-source way, that other places could use.”

The tool, dubbed TILE2NET, was developed using a few US territories as initial data sources, but can be refined and customized for use anywhere.

“We thought we needed a method that could be scaled and used in different cities,” says Maryam Hosseini, a postdoc in MIT’s City Form Lab in the Department of Urban Studies and Planning (DUSP), whose research focuses heavily on development from the tool.

The paper “Mapping the Walk: A Scalable Computer Vision Approach for Generating Sidewalk Network Datasets from Aerial Imagery” is published online in the journal Computers, Environment and Urban Systems. The authors are Hosseini; Sevtsuk, Charles and Ann Spaulding Career Development Associate Professor of Urban Science and Planning at DUSP and Director of MIT’s City Form Lab; Fabio Miranda, assistant professor of computer science at the University of Illinois at Chicago; Roberto M. Cesar, Professor of Computer Science at the University of Sao Paulo; and Claudio T. Silva, associate professor of computer science and engineering at New York University’s (NYU) Tandon School of Engineering and professor of data science at the NYU Center for Data Science.

Key research for the project was conducted at NYU when Hosseini was a student there, working with Silva as a co-adviser.

There are several ways to attempt to map sidewalks and other pedestrian pathways in cities and towns. Planners could create maps manually, which is accurate but time-consuming; or they could use streets and make assumptions about the extent of sidewalks, which would reduce accuracy; or they could attempt to track pedestrians, which would likely have limited ability to display the full range of pedestrian networks.

Instead, the research team used computer-aided image recognition techniques to build a tool that visually recognizes sidewalks, crosswalks and footpaths. To do this, the researchers first used 20,000 aerial photos from Boston, Cambridge, New York City and Washington – places where comprehensive pedestrian maps already existed. By training the image recognition model on such well-defined objects and using parts of those cities as a starting point, they were able to see how well TILE2NET would work elsewhere in those cities.

Ultimately, the tool worked well, recognizing 90 percent or more of all sidewalks and crosswalks in Boston and Cambridge, for example. After the tool has been visually trained on these cities, it can be applied to other metropolitan areas; People elsewhere can now also feed their aerial photos into TILE2NET.

“We wanted to make it easier for cities in different parts of the world to do something like this without having to do the heavy training [the tool]’ says Hosseini. “Hopefully, together we’ll do better and better as time goes on.”

The need for such a tool is enormous, points out Sevtsuk, whose research focuses on pedestrian and non-motorized movement in cities and who has developed several types of pedestrian mapping tools in his career. Most cities have incomplete networks of sidewalks and walkways for pedestrians, he notes. And yet it is difficult to efficiently extend these networks without mapping them.

“Imagine if we had the same gaps in car networks as pedestrians have in their networks,” says Sevtsuk. “You would drive to an intersection and then the road just ends. Or you can’t turn right because there is no road. That’s what [pedestrians] are constantly confronting us and we don’t realize how important continuity is to us [pedestrian] networks.”

In an even larger context, according to Sevtsuk, the ongoing climate change means that cities, among other things, have to expand their infrastructure for pedestrians and cyclists; Transport remains a huge source of carbon emissions.

“When cities talk about reducing carbon emissions, there’s no other way to make a big dent than to address transportation,” says Sevtsuk. “The whole world of city data for public transport and pedestrians and bikes is really far behind [vehicle data] in quality. Such data is needed to analyze how cities can function without cars.”

On the plus side, Sevtsuk suggests adding pedestrian and bicycle infrastructure “will be done more aggressively than in many decades in the past. In the 20th century it was the other way around, we removed sidewalks to make way for vehicular roads. We are now seeing the opposite trend. In order to make the best use of pedestrian infrastructure, it is important that cities have the network data about it. Now you can really tell how someone gets to a bus stop.”