Sign In

Communications of the ACM

ACM TechNews

Personalized 'Deep Learning' Equips Robots for Autism Therapy


A therapy session augmented with the NAO humanoid robot, showing how facial and limb motions were tracked.

Massachusetts Institute of Technology Media Lab researchers have developed a deep learning network to help robots gauge the effects of autism therapy on children.

Credit: MIT Media Lab

Researchers in the Massachusetts Institute of Technology (MIT) Media Lab have developed a personalized deep learning network to help robots gauge the effect of autism therapy on children, using data that is unique to each patient.

The research team used humanoid NAO robots that exhibit emotional states via eye color, limb movement, and tone of voice, while interacting with 35 autistic children in 35-minute-long sessions.

MIT's Oggi Rudovic says most subjects responded to the robot "not just as a toy, but related to NAO respectfully as it if was a real person," especially during storytelling.

The personalized deep learning framework could learn from video, audio, and physiological readings collected on each child, while information about the child's autism diagnosis and abilities, their culture, and their gender also was considered.

The robots' perception of the subjects' responses correlated 60% with human experts' assessments.

From MIT News
View Full Article

 

Abstracts Copyright © 2018 Information Inc., Bethesda, Maryland, USA


 

No entries found

Sign In for Full Access
» Forgot Password? » Create an ACM Web Account