Studying how kids interact with robots
A visiting scholar probes human-robot interaction and how children on the spectrum react
Lundy Lewis, an academic and researcher in artificial intelligence and human-robot interaction, is in Kanata watching a pair of sixyear-old boys playing with social robots in the gym at CHEO’s site for autism.
Griffin and James Beck are twins. The robot they’re interacting with is called Jibo, developed at the Massachusetts Institute of Technology. Jibo has no arms or legs and only two joints, one which approximates a neck and another a waist. Despite this, Jibo can pack a lot of emotion into his rotund body. Equipped with facial recognition and a touch screen, Jibo turns his head toward people. He can appear concerned, merry, quizzical.
Jibo responds to the twins’ request to dance, playing a tune and gyrating comically. The boys ask him to dance again and again.
“You should see Jibo twerk,” notes Lewis, the Fulbright visiting research chair in science and society who is in Ottawa to conduct four research projects that look at children’s interaction with social robots.
Children with autism sometimes have a difficult time interacting with humans. “We have millions of muscles in our faces we’re not even aware of. They have a hard time following that,” said Lewis, who has also researched robot interactions with dementia patients.
Lewis’ study at CHEO is looking at how nine children on the autism spectrum interact with a variety of different robots compared to their favourite toys. Typically, autistic children can be motivated by interacting with the toys, but they can also be engaged though affectionate exchanges — a high-five, for example — or edible rewards. The question Lewis wants to answer is whether social robots offer additional advantages to the existing methods of interacting.
“Everybody seems to like robots. Why? We’re always making representations of ourselves,” he said. “We love plays. We love to see ourselves acted out on the stage, or in movie. It’s the same with robots. We like machines with our likeness.”
Children with autism can practise role-playing with social robots. Robots don’t show disappointment, frustration or boredom. Children can pick up on these emotions, he said.
Only one of the robots in Lewis’ study looks like a human. The robots include a cat, a dog and a baby seal named PARO developed in Japan that has already received a lot of attention for its potential therapeutic uses. There’s also a humanoid robot developed in France called Noa, which Lewis believes has potential for children with autism. But there are also problems, like fingers pinched in Noa’s joints.
Cailin Hudson, the clinical coordinator of CHEO’s autism program and Lewis’ co-investigator through CHEO, said robots have potential for teaching more independence and expanding social exchanges. Jibo, for example, has a number of greetings, ranging from “Top of the morning!” to “Hey! How are you?” Children with autism can learn that these are both acceptable and can be used in similar contexts.
Children on the autism spectrum are all different. The twins’ mother, Tiffany McCunn, said James is the more social and verbal twin. He likes cars and trains. Griffin is more drawn to technology and already has a robot toy called Cogsley that plays music, dances and teaches numbers and letters. While Griffin enjoys Cogsley, James is not particularly interested.
Cogsley is a simple and fun way for Griffin to engage. He stays focused and learns well, she said. “A lot of kids on the spectrum are attracted to technology. It helps bridge the gap to social interaction.”
Lewis says there are so many research questions that have not been answered yet.
“We have the technology, the conceptual paradigms and the engineering to do anything we want. Simple ideas are the best.”