For children growing up with autism spectrum disorders (ASD), interacting successfully with others often presents enormous challenges.
When it’s customary to make eye contact, children with ASD often gaze elsewhere. When a smile would be appropriate, they may deliver a scowl. And when a playmate communicates frustration via a facial expression, autistic children often don’t recognize the signal, responding with behavior that makes matters worse.
As a result, children with ASD often struggle to make friends or thrive in any setting that requires people skills.
Enter NAO, a personable robot with a mission. He’s helping an interdisciplinary faculty-student research team, based out of DU’s Daniel Felix Ritchie School of Engineering and Computer Science, conduct a pilot study exploring whether humanoid robots can improve social and communication skills in children with ASD.
“He can walk, talk and dance,” says Mohammad Mahoor, assistant professor of electrical and computer engineering. The 23-inch-tall NAO, made by Aldebaran Robotics of France, can also direct autistic children in a host of activities designed to improve their recognition of facial expressions and to help them direct their gaze appropriately. When the kids succeed, NAO can even enlist them in a celebratory high-five.
The robot project is one of several research initiatives led by Mahoor, an expert in visual pattern recognition, social robot design and bioengineering. With this project, he aims to build on studies pointing to the therapeutic potential of robots for the ASD population. He hopes to enlist up to 50 children with high-functioning autism in his two-year study, scheduled to conclude in another year if funding permits.
“You may ask, why a robot? Why not a human?” Mahoor says of his research premise. “Humans are very overwhelming for kids with autism.”
Toys and technology, on the other hand, are less intimidating. “A lot of kids on the spectrum like mechanical things,” says Sophia Silver, a junior psychology major intent on a career in child psychology. Silver brings her insight into autistic children to the research team. She also recruits study participants, schedules their visits and talks with parents while their children are playing with NAO.
To date, Mahoor’s study has enlisted 24 participants, ages 7 to 17, who, over the course of six months, come to a University experiment room every two weeks for 30-minute sessions with NAO. Equipped with four microphones and two cameras, NAO records lots of data about each participant — everything from the duration and frequency of their direct gazes to the range of their facial expressions.
“Why are these important?” Mahoor asks. “These are the bases of human sociability.”
NAO is programmed, scripted and operated by members of the DU research team. Huanghao Feng, a graduate student in computer engineering, has been with the project from the beginning. Initially, he joined the team to learn more about the newfangled NAO, but the project has also taught him a lot about working with autistic children. When they don’t respond to the robot, or when their attention wavers, he repeats the robot’s requests or helps them stay on task.
One of the activities enlists a handful of small beanbags, each of them sporting a photo of a person demonstrating an expression — happiness, perhaps, or sadness or anger. The child is asked to find, and show NAO, the toy with the angry face. This exercise helps participants identify the emotions attached to facial expressions—a skill, Silver says, that often eludes children with ASD.
“If kids can’t identify that another child is angry, they’ll get in more fights. They’ll have trouble making friends,” she says.
Other exercises, Mahoor notes, work on what is known as joint attention—in other words, shared focus on an object. NAO may, for example, ask the child to follow his gaze to, say, a line of boxes. “I have kids who are able to follow what NAO asks them to do. ‘Look at that object or pick up that object.’ Then NAO gives them a hug or a candy, a reward,” he explains.
These moments of triumph are captured on NAO’s cameras. “After the sessions,” Feng says, “I go back to the lab and process all the data.”
The cameras capture some interesting information, such as the duration of direct gazes and the frequency of gaze shifts. Preliminary findings suggest that NAO is helping some of the children maintain a direct gaze for longer periods.
Mahoor and the research team are hoping to add a second robot, Zeno, into the mix. Made by Hanson Robokind of Dallas, Zeno can display a range of facial expressions and instruct participants in mimicking them. And, Feng adds, like NAO, he can record those expressions, allowing the research team to gauge the social development of study participants.
Robots bring another advantage to a research team intent on repeating exercises and measuring data over time, Mahoor says. Robot interactions can be conducted precisely the same way again and again. That simply wouldn’t be possible with a human, who might introduce a new variable into a game or conversation.
It’s too soon to know just how effective robots can be. Not every child engages with them, but Feng and Silver have witnessed significant progress in several children.
Take the case of one nonverbal little boy. “At the beginning,” Feng says, “he was completely scared of the robot. But after several sessions, he hugged the robot and kissed the robot. He even came up and hugged me.”
Silver takes equal pleasure in such developments. “To see them excited about something is fun,” she says. “And it is really liberating for their parents as well.”