DU Professor Uses New Techniques to Treat Autism
By: Marcus Turner
Dr. Mohammad Mahoor, assistant professor of Computer and Electrical Engineering at the University of Denver, is making tremendous strides in the treatment of children with autism. Though the department of engineering may not seem like a typical place to perform this research, Dr. Mahoor has found a way to utilize face mapping technology, robotics and artificial intelligence to assist children on the autism spectrum.
"Robots act as an agent to help children improve social behaviors," says Mahoor. "Autistic children are very interested in toys, and the robots help create a connection that fosters social health." The children, ages 7-17, interact with different human-like robots to refine skills such as eye-gaze attention. Mahoor has seen an improvement in children's attention after working with his robots for a short duration of six months. These "companion bots" are seen almost as a pet to the children. "Except a pet that can talk," says Mahoor.
In some sessions, the robot actually holds a conversation with the child. "The children come for twenty to thirty minutes and play a series of games," Mahoor explained. "For example, the robot asks the child to find a box in the room that is marked with a specific expression like a smile or a frown." Performing this simple task, the children recognize different emotional expressions and are having fun at the same time. In these same sessions, the robot asks to hear a story from the child. "The child will then tell the robot a story," Mahoor said. "The robot then responds to the story and the child is interacting in a socially healthy way."
Healthy social interaction is not only an important step in the treatment of autism but also in the more common affliction of depression. "[Robots provide] proactive dialogue on depression," Mahoor said. "For example, if I have a pessimistic behavior, that robot can help me improve by reaffirming me." Specifically, Mahoor has plans to use his robots to treat depression in elderly patients who may spend large amounts of time alone. "If something bad happens, the robot can recognize the event, recognize the pessimistic tone of the speech and console and comfort the subject," said Mahoor.
Another factor that Mahoor considers is the appearance of the companion-bot. "Robots are often cartoonish and friendly so the subject may have an easier time interacting," he explained. Mahoor also explained that when a robot has a human appearance, the patient is "creeped out" and does not want to interact. "The look and appearance are very important in the design," he said. He gave an example of a Japanese robotics company that has experienced great success with therapy robots designed to look like baby seals.
Currently, Mahoor and his students are working to build two robots for these treatment sessions. One resembles a large teddy bear in the body but its mouth is composed of computer graphics. "We wanted the mouth to create human expressions so we created a projection system that projects the mouth onto a screen," he explained. Having more options for moving the mouth of the robot creates more opportunities for the patients to engage and mimic what they see. Mahoor expects this robot to be completed within this year.