Smartphone Movements Could Reveal Empty Parking Spots
Researchers say “pocketsourcing” could let you find parking spots easily, without requiring cities to add spot sensors.
Cities are trying to find new ways to keep track of parking availability and reduce congestion.
Researchers have come up with a novel way to find parking spots with your smartphone. It promises to be much easier than driving around looking for an empty space, and doesn’t require the installation of pricey sensors or other methods for tracking available spots.
At the State University of New York at Buffalo, researchers built an app called PocketParker that does what they’re calling “pocketsourcing”—essentially, turning smartphones into passive sensors that track the location and movements of other users who’ve installed the app. A remote computer crunches the aggregate user actions and determines the likelihood that a lot has an open space. A paper about PocketParker will be presented at the ubiquitous computing conference UbiComp in Seattle this week.
While some parking lots employ sensors to gather information about capacity, PocketParker works without any such infrastructure. It pulls parking lot data from OpenStreetMap and calculates the number of spaces in a given lot based on its dimensions. During a study, researchers found that they could predict the number of spaces to within 6 percent of the actual number.
The app uses the smartphone’s accelerometer to determine where a user is and gauges whether he’s looking for a parking spot based on his movements. If a user drives slowly through a parking lot without stopping, that signals that the lot is full. If a user displays movements typical of walking and then suddenly speeds up and leaves the lot, that signifies that he likely just got into his car and drove away. The app calculates this in the background. “There should be no interaction required,” says SUNY Buffalo computer science professor and paper coauthor Geoffrey Challen.
For their study, the researchers had 105 smartphones users around Buffalo test out the app over a month and a half, generating a total of 10,827 car arrivals and departures. Checking their work with cameras they installed at the lots they tracked in the study, the researchers found they were able to correctly predict how many spaces were available 19 out of 20 times. “Our goal is to prevent people from circling,” Challen says.
There are a few problems with the approach. One obvious one is that PocketParker can’t account for drivers who aren’t using the app. Another is that a user might leave his phone in the car, drop his car off and get in another car, or he might not be searching for a parking space at all but be picking up a friend. “Until you have enough people using it, apps like this tend not to work well,” Challen says. “It’s stuck in this chicken-and-the-egg problem.”
Because of this, Challen doesn’t envision PocketParker as a standalone app. Instead, it could be featured within a mapping app—similar to the way Google Maps integrates traffic data into its app. Challen believes that if a feature like PocketParker were flipped on in the background, it would quickly collect enough data to make far better assumptions about parking spot availability.
Keep up with the latest in intelligent machines at EmTech Digital.
The Countdown has begun.
March 25-26, 2019
San Francisco, CA