Google Street View and crowdsourcing could soon fix potholes

Project Sidewalk
Credit: New Scientist

Every neighborhood has them - that gaping pothole that threatens to eat your front end, or the sidewalk that is no match for growing tree roots. Getting these problems taken care of quickly is the idea behind a new software program created at the University of Maryland, using crowdsourced workers and Google Street View. When a problem is spotted a report can quickly be sent to the responsible city council for targeted attention.

One driving force behind the development of the prototype was to make neighborhoods easier for the handicapped to navigate safely. In fact, three wheelchair users were used in the testing of the software, along three ambulatory users. Together they viewed a video of how to spot and label problems with identifying colored shapes. The test then went to 400 crowdsourced workers from Amazon's Mechanical Turk Service. The result was a 93 percent success rate in identifying problems.

Jon Froehlich and researchers at the University of Maryland in College Park created the software after the idea generated interest from the Department of Transportation (DOT). The DOT believes the project could save local councils and townships time and money by being able to direct them exactly where a problem lies, along with information on what it is.

Armed with advance information teams can be scheduled, staffed and given the proper resources more efficiently before heading out on a job.

The researchers are already looking forward to advancing the prototype to use computer  vision algorithms and access the laser range-finding data the Google Street View cars record as they cruise through the streets. The automated tools would ensure even greater accuracy in identifying problems, and hopefully save municipalities money on road works. While there are existing apps and websites that encourage people to reort problems when they see them, this software streamlines specific data for local officials using data that already exists and is constantly updated.

Froehlich will present the software prototype in April at the Conference on Human Factors in Computing Systems in Paris, France.

Via New Scientist