Connectivity

A phone that says “no” to little kid fingers

An algorithm for detecting who’s swiping the screen could help make phones child-proof.

It may soon be possible for your phone to automatically figure out whether it’s you or your five-year-old who’s swiping the screen—and, if it’s the latter, block apps you want to keep off-limits to kids.

That’s the vision of researchers at the University of South Carolina and China’s Zhejiang University, who’ve created an algorithm that can spot whether your kid is accidentally trying to, say, order from Amazon without your knowing.

There are already plenty of activity-monitoring apps that aim to control what kids do on phones, but parents need to add them and turn them on, and they could be disabled by tech-savvy children. The researchers figured that automated age-range detection would make it easier for parents to hand their phones over to curious children without worrying that the kids will stumble upon an inappropriate website or get into a work e-mail account.

Xiaopeng Li, a graduate student at the University of South Carolina, coauthored a paper on the work that will be presented at a mobile tech conference next week. He says the researchers observed two big differences between how children and adults swipe phone screens.

Since kids have smaller hands and shorter fingertips than adults, they often touch a smaller area on the screen and make shorter swipes. Children also tend to swipe their fingers more sluggishly across the screen, and they are slower to switch from swiping to tapping.

To get hard data on these differences, the researchers built a simple app and asked a group of kids between the ages of three and 11—and a group of adults between 22 and 60—to use it. The app had participants unlock an Android phone and then play a numbers-based game on it, so that the researchers could record a variety of taps and swipes. They also tracked things like the amount of pressure applied by a user’s finger and the area it encompassed.

The researchers used the resulting data to train an age-detecting algorithm that they say is 84 percent accurate with just one swipe on the screen—a figure that goes up to 97 percent after eight swipes.

To make the approach even more effective, Li says, the team wants to incorporate indicators such as a user’s movements (trackable using a smartphone’s accelerometer), since the researchers also observed that kids’ hands seemed to shake more than adults’ when holding phones. The algorithm hasn’t been built into a phone yet, but it looks like a really promising way to ensure that little fingers don’t tap in the wrong places.