Rutgers students developing system to provide parking availability information

"People have counted how long cars circle around these blocks, and there are some amazing statistics that have come up," said Gruteser, an associate professor of electrical and computer engineering. "In the Park Slope area of Brooklyn, up to 45 percent of the cars driving seem to be cruising for parking."

Gruteser, his students and Wade Trappe, associate director of the Rutgers WINLAB, were looking into real-time traffic monitoring when they realized there were limitations.

Tracking the speed of mobile phones, they could estimate traffic flow and how long it would take to get to Manhattan. But then there was the next step: How long does it take to find a parking spot?

"At some point, we realized that theres this other story once you get to your destination, you need to find a place to park," Gruteser said. "And getting real-time parking information is difficult and there wasnt any good source for it. So we came up with this way to sense parking information and be able to get it to your navigation system."

Helping drivers in congested areas

Using ultrasonic sensors, GPS location finders and wireless networks, the Rutgers group thinks it has come up with a cost-effective way to locate the nearest open parking spaces and give drivers choices.

Taxis, municipal vehicles or mall security cars vehicles that travel frequently in a given area would be equipped with sensors that measure distances to obstacles and determine whether there is an available parking space. That information would then be fed to an internet server and matched with a map of legal parking spots.

Drivers would then be able to find out, through their GPS navigation systems or smart phones, where parking spaces are open, saving them the time and frustration of looking for a spot.

Using algorithms, the Rutgers team was able to distinguish between parked cars and other objects such as trees or fire hydrants.

"There are some other obstacles that can be on the side of the road that have that same size and eventually we learn over time, if this obstacle never moves, its always there, its probably not a car," Gruteser said. "If this space is always open and in an area where parking is very crowded, its probably not a legal spot."

The project started two years ago, and researchers quickly were able to achieve 90-percent accuracy. Students attached sensors in a case resembling a Tupperware container outside their vehicles and collected data on parking spots in Highland Park and Brooklyn at under 40 mph to get more accurate readings.

The future of parking

Gruteser said he hopes the product will be ready for widespread use in about two years.

Too bad it wont be ready in time for this years holiday shopping season at New Jersey malls, where parking lots can get more crowded than the stage at an Earth, Wind and Fire concert.

Gruteser said a ParkNet sensor would probably cost a few hundred dollars per vehicle if rolled out today, but the price could eventually be brought down. He said drivers would pay just a few dollars to get the parking information on their navigation system.

Steve Carrellas, the New Jersey representative of the National Motorists Association driving-rights group, was encouraged the system would be inexpensive and not involve digging into parking spaces to install sensors.

"The knowledge of an available space can be valuable," said Carrellas, himself a professional engineer. "You dont have to worry about distractions as well" such as trying to keep one eye on the road and another looking for a parking spot.

He also saw commercial applications for towns wanting to better manage their parking.

The next step is to find a company willing to test the device, preferably in a congested area like New York City.

As part of a $20 million project, San Francisco has started putting sensors under downtown parking spots that can detect when a car parks on it. Gruteser said ParkNet is a more cost-effective alternative.

Student Tong Jin, who is doing his masters thesis project on the ParkNet research, gave a recent demonstration of how the system works by attaching a sensor to his car and driving a couple loops around a Rutgers parking lot.

The system would be most beneficial in, but is not limited to, the largest cities.

"Think of areas like Princeton its a small town, but in the downtown area, parking is very crowded," Gruteser said. "Pretty much any small city or central business district or Main Street area where parking is crowded could benefit. Certainly areas like Hoboken."

Or the data could come in handy for towns making decisions on how to divvy up parking spaces.

"City authorities might be interested in the raw statistics of how many spots are there at different times of days to figure out, where do we need additional parking meters, where should we change the rules?" Gruteser said. "Theres a lot of parking decisions they have to make that could be based more on actual data, rather than intuition."