CNN - Intelligent playgrounds
September 26, 2007

LONDON, England -- Pick me! Pick me! The weakest children may no longer be left out of playground games. New technology may help to put kids on a more level playing field, which may in turn motivate them to learn and encourage competitiveness. Using modern artificial intelligence and robotics, new playground games can recognize a child's behavior and respond accordingly -- in real-time -- to make the game harder or easier.

The industry calls it augmented cognition, or 'aug cog', a technology that is also being developed by the armed services to reduce mental overload in the battlefield. For example, fighter pilots helmets can be equipped with sensors to distinguish when the brain is becoming overdosed. When that happens, a computer will adjust the level of incoming stimuli, dimming the interface and lowering the volume of messages. The end goal is that a computer will eventually be able to judge whether incoming material is important enough to interrupt your current activity.

The research aims to create more effective military personnel but it could be used by anyone who has to cope with multiple information streams, for example stock brokers under stress. Aug cog is also being studied for its applications in the gaming world. It can be utilized in video games to raise or lower difficulty levels, thereby ensuring that a player is sufficiently stimulated but not overwhelmed.

It may be children who are experiencing the benefits of aug cog first. Researchers at the University of Southern Denmark at Odense have built special playgrounds at two schools, two kindergartens and two youth clubs in the city. Henrik Hautop Lund, professor of robotics, says he wants to transform playgrounds for the new millennium. "It has been incredible to see how immediate children respond to them," Lund says. "They immediately took to this technology."

Lund and his team came up with 'intelligent tiles', a grid of pressure-sensitive floor tiles for playgrounds. Each tile has a small computer inside it, which measures the force of the child's foot and responds with either colored lights or sounds. One game, called 'Bug smasher', provokes children into chasing lit 'bugs'. When a bug lights up, the child must tread upon the surface of the tile and smash the bug. They are rewarded by a comical, smashing sound.

What makes this game different is that it adapts to the abilities of different children. If one child is slower than another, the game will react accordingly, for example, slowing down the speed of the tiles lighting up. The playground will operate more quickly for a youthful, agile user and slow down for say, an obese child. "The playground itself automatically figures out how to change its game behavior," Lund says.

The team at the University of Southern Denmark developed the technology by first studying children in a playground. They categorized the behavior of children, comparing those who played in a disruptive manner with those who played in a continuous way. When they brought a new set of children to the playground, the neural network they had programmed had learnt to recognize different children's abilities. It could even distinguish when a child was tiring. Every thirty seconds, the neural network re-categorized the child and changed its response if necessary.

"For example, if a child isn't playing the game at all, the playground moves the 'intelligent tiles' closer to the child," Lund says. "The child may touch one of the tiles, almost by accident. When they become aware of the smashing sound, they may think 'what's that?' and start to play the game."

Lund found the neural network was able to characterize children correctly 97 percent of the time. "The playground should adapt to each particular child," Lund says. "If it's a slow obese child, we should make it easier to play the game in order to encourage the child to do physical activity."

The technology can also make for more interesting playtime for parents, too. In another game, called 'ping pong', a parent and child hit a virtual ball between each other in the playground. Instead of the parent having to slow down their game to suit the skills of the child, the technology automatically made it equally challenging for both sides. The virtual ball became bigger for the child and slowed down. For the mother, the game was quicker and the ball shrank.

"The game aims to operate at 10 percent above the capabilities of both the parent and child, so they are both challenged at the same level," Lund says. "It actually becomes interesting for the adult to go the playground. Often parents go there just to watch and wait for their children finish on a swing. Children can feel that pressure of their parents, too, so it's better for everyone."

Denise Nicholson, Professor of Modeling and Simulation at the University of Central Florida, is also researching aug cog in the gaming industry, as well as in education and even advertising. "We want to understand more about people's reactions and find ways to measure that."

Nicholson is currently looking at a system, which will aid speech therapy. The first phase is to give speech therapists the tools to play different songs to students, who must respond, for example, by repeating something softly or loudly. The system will measure the student's pitch, tone and volume, and furnish the therapist with a spectrum analysis of the session.

"Using aug cog, the end goal is to have a system which would have a feedback-adapting scenario," Nicholson says. "It's one thing to make the system sense what the person is doing but quite another if you want to make it adapt and be effective. In reality, there is still much research going into what is an effective adaptation."

Nicholson says the long-term goal is richer, real-time adaptation. That will require multiple sensors, like measuring heart rate and electrical activity of the brain, as well as eye-tracking. "We still need to find out which sensors are most reliable but I think within the next few years, we'll see practical implementations in the market place. Systems will just use one or two sensors but it will be a beginning."

Lund is also working on other applications, including 'therapy tiles' for use in physiotherapy, sports training and possible diagnosis of children with autism. Together with Entertainment Robotics, they are already helping cardiac patients recover. "We are pretty much doing the same things as with children," Lund says. "We are making the same kinds of tiles a little more sophisticated. Therapists put them up on a wall and construct social games where recovering patients can increase their pulse rate by playing against each other. Rehab can be repetitive and dull but the feedback we are getting is that this is very motivating."