This year at the FIFA World Cup, a computer will know the second a shot passes the goal line—sometimes even before the referee can make the call. The system, called GoalControl, uses a set of 14 high-speed cameras to monitor each goal and send a signal to a smartwatch any time a ball breaches the goal line.
The system is just as—if not more—accurate than a human referee. It completed a successful trial at last year’s Confederations Cup, in which it correctly scored all 68 goals scored in the tournament.
So sports fans (especially those whose teams have been eliminated or who just aren't into soccer) are left wondering: Can the same level of automation be applied to other sports—perhaps even American football?
That’s the question a team of North Carolina State University and Carnegie Mellon researchers recently set out to answer. The group, funded by Disney Research, shared their system, which uses radio transmissions to track a football in 3D space, in IEEE Antennas and Propagation Magazine earlier this month.
From the onset, the team knew that a system like FIFA’s camera-based ball tracking wouldn’t work for them.
“American football is unique, because the ball is obscured a lot—particularly when you care about it the most,” says David Ricketts, an associate professor of electrical and computer engineering at North Carolina State University and one of the paper’s authors. Players tackle and pile up on the ball, because that’s how the game is played.
The team embedded a low-frequency radio transmitter and small battery pack into the football; a series of antennas track the signal, and a computer uses that data to map the ball’s position in 3D space. The lower frequency signal produces wavelengths that are extremely long, which means they can easily pass through the human body.
According to Ricketts, frequency interference is what went wrong with some past ball-tracking attempts. In those cases, engineers used transmitters that produce high-frequency waves, which are easily absorbed by the human body—not exactly the best way to get a signal from underneath six linebackers.
But low-frequency waves aren’t perfect, either: The ground absorbs them and then re-emits and distorts them. “Even though people are transparent,” Ricketts says, “the Earth is not because it’s just so big.”
While their approach on the whole might be new, the team used an old method to unscramble the signals.
“We realized that we could use a technique developed in the 1960s called complex image theory,” said Dr. Darmindra Arumugam, another author and a former Ph.D. student at Carnegie Mellon, in a press release. “Complex image theory allows us to account for the secondary fields generated by the earth and compensate for them in our model.”
Even with the transmitters onboard, footballs will still adhere to NFL regulations. “A football has laces, which make it very asymmetric, variable and lopsided,” Ricketts explains. The NFL also allows for a weight variance of up to a half an ounce; the teams transmitters weighs less than 25 grams.
Right now, the system is accurate within about 77 centimeters; it needs to get down to 14 cm—about half the length of a football—in order to be a viable option for game play.
While the technology certainly won't be ready for the upcoming season or even the next, Ricketts isn’t discouraged.
“Trying to locate a football on the field is like trying to find a one-foot cube in a 450,000-cubic-foot-area,” he says. “It’s not like there’s a fundamental limitation in the engineering—it’s really about improving the algorithm that we use to estimate the position.”
While the system might be helpful for handling disputed calls, establishing first downs, and lines of scrimmage, Ricketts thinks it can also help enhance the TV viewing experience in the same way MLB strike-zone cameras or NHL puck-tracking have already.
“If you think about it,” he says, “when everyone’s lined up and the center’s in place, you don’t even see the ball until he hikes it.”
The next likely step: a hand off of the technology to a corporate partner for further development and potential implementation.
“It’s ready for a technology transfer,” Ricketts says.