Street Bump uses the phone's accelerometer to detect bumps in the road. When a car hits a pothole, the app sends information about the bump, including its location, to a database.
But the original version of Street Bump couldn't distinguish between potholes, manhole covers, bridge surfaces, and various other road lumps. As a result, it was giving a lot of false positives.
In the spring of 2011, Boston officials sought the help of InnoCentive, a firm that uses crowdsourcing contests, or "Challenges," to help its clients find solution to tricky technical problems.
[ Learn more about big data. Read 10 Big Predictions About Big Data. ]
"They came to us and said, 'OK, we want to make this thing better. We want to get it ready for prime time,'" InnoCentive VP of Marketing Steve Bonadio told InformationWeek. "'We want to create a Challenge, put up the prototype (app) and data points for people to evaluate, and have them come up with better algorithms and mechanisms for detecting actual potholes.'"
The resulting InnoCentive contest drew more than 700 "solvers"--twice the usual number of Challenge participants--who submitted 19 potential solutions. Liberty Mutual Insurance put up $25,000 in prize money for the winners.
"We worked with the City of Boston's New Urban Mechanics group to identify three winners, two of whom were awarded $9,000 each, and one was awarded $7,000," Bonadio said.
The final algorithm included elements from three winning solutions, including one from Edward Aboufadel, a professor of mathematics at Grand Valley State University in Michigan. His approach included the use of wavelets, Kruskal's algorithm, and other mathematical tools to accurately identify potholes.
A second winner came from Sprout & Co., a non-profit group in Somerville, Mass. It involved the use of magnitude-of-acceleration spikes along the phone's z-axis to spot impacts, and additional filters to distinguish potholes from other irregularities on the road.
The new algorithm made Street Bump, a free download in Apple's App Store, a winner.
"Simply by driving our streets and running this app, our Public Works Department will be getting information it can use to both dispatch repair crews and prepare long-term capital plans," said Boston Mayor Thomas M. Menino in a statement.
An Android version of Street Bump is under development, but currently isn't available in the Google Play store.
The City of Boston plans to make the Street Bump algorithm available to other municipalities, and has received interest from cities around the world.
Other cities "want to use the source code and algorithm behind this to create similar apps in their regions," Bonadio said.
InnoCentive's crowdsourcing approach to solving technical problems is fairly new. The company was founded within pharmaceutical giant Eli Lilly in 2001 to find novel ways to solve problems outside of the traditional R&D process. It was spun off as a private company in 2005.
Innocentive invites its community of 260,000 people, who sign up for free at the company's website, to develop solutions to its clients' problems.
"We help evaluate the solutions, select one or more winners, and award solvers a monetary award," said Bonadio. The Street Bump Challenge is a "very good example of using an innovative tool like crowdsourcing to reach out to a diverse group."