Avi Cooper doesn’t want you to get lost in the mall.
Sure, there are always the maps — assuming you’re not lost inside a department store with poor signage.
But Avi, 17, is a senior at the Torah Academy of Bergen County in his hometown, Teaneck. He is of the generation that has grown up taking GPS for granted. If your phone can help you navigate the towns of New Jersey, why can’t it help inside the Garden State Plaza?
GPS doesn’t work indoors for two reasons. The satellite radio signals that GPS uses don’t penetrate well into buildings. And “GPS is not very precise,” Avi says. “At best it has five meters accuracy.” That enough to know what street you’re driving on, but not which hallway you’re walking down.
Avi, though, brings us good news. He spent the summer working on the problem of indoor GPS in a Rutgers lab, along with Poojit Hegde of Edison, a high-school junior at the Middlesex County Academy for Science, Mathematics, and Engineering Technologies. The two found their way to Rutgers through Partners in Science, a Liberty Science Center program.
The approach they took for solving the GPS in the mall problem is summarized by the paper they wrote on the topic: “An Indoor Positioning System Facilitated by Computer Vision.”
The two teens recently presented their paper at the Massachusetts Institute of Technology. It came in second among the 35 papers accepted by a technology conference sponsored by the MIT chapter of the Institute of Electrical and Electronics Engineers. That’s not bad for two high school students competing against college students.
Actually, Avi left the formal delivery of his paper to Poojit, because the presentation was scheduled for Shabbat. He did get special permission to present a poster about the project on Sunday.
The solution to the GPS problem was to have each room track people as they walk through it. “We set up stereoscopic cameras to act like eyes,” Avi said.
With two cameras in a room, Avi and Poojit then researched “the different ways a computer can identify somebody and position them,” Avi said. “By the end of the summer we got our system working to the point where it could recognize one of us with 95 percent accuracy and to within a foot.”
They used a free software library written by Intel, which handled the basics of recognizing people.
“We figured out where people are in the room through background subtraction,” he said. “If we remove the stationary pixels, we can assume the only thing moving in a room is a person.”
From there, the computer figures out what the person looks like. That’s important so it can know which one is you, and track you. The program Avi and Poojit wrote keeps tracking a person, even as the image changes directions and apparent size to the cameras.
Avi thinks that this technology could be useful to shoppers and store owners.
“If you’re in a mall or an airport, you could get walking directions on your phone to a gate or a store,” he said. The technology also would give the owner of the mall or airport more data about how people use the facility, he said.
Avi also has worked on a program with more potential immediate benefits for his classmates: “a social network/sports reporting app for the yeshiva high school sports teams. It’s a work in process.”
Last year, it earned him a scholarship from Apple to attend its developers conference in San Francisco.
Avi has not studied computer science at school. “Everything I know is stuff I learned from the internet,” he said.
For college, “I’m probably looking to major in computer science and do more research,” he said. “I really enjoy not only learning stuff in the classroom, but going out and doing things and seeing what my capabilities are.”