• ICare4Autism

Researchers Use A.I.-Based Intervention to Help Children With Autism

Everyday social interactions can pose a major challenge for many children with autism spectrum disorder. In a new study, published Wednesday in Science Robotics, researchers reported that robots, programmed with artificial intelligence, can be used to help children with autism improve the social skills needed to communicate more effectively. In their study, researchers used Socially Assistive Robotics (SAR), a new computer model that offers personalized visual and verbal feedback to children with autism as they play space-themed math games on a tablet. The robots use these games to prompt the children to perform social tasks. The researchers also analyzed video and audio recordings collected through robots that were tested in the homes of seven children with ASD. According to a report by, the computer model had a 90% accuracy rate in recognizing when to provide the children with feedback and information about a task. “Children with autism face developmental delays, especially when it comes to their social skills,” Shomik Jain, a machine learning scientist at the University of California and an author of the study, said. “There’s been a lot of research in showing the potential for robots to help develop these social skills in children. When specific games are designed that teach children to take turns or speak out loud, or when children are required to verbally count and do basic arithmetic, as in our study, we find that just by listening to the robot give instructions and feedback and responding to those instructions, the child improves overall in his or her ability of social interaction.” Jain explained that the robotic system used by the researchers consists of two parts. “One is obviously the robot, a small bird-shaped robot that is hopefully appealing to the children,” he said. “We also [have] a tablet for [the children] to play math games. The robot would provide the instructions and feedback about the task, while the child would play the games on the tablet.” Jain clarified that SAR is not a replacement for therapy, but can augment therapeutic intervention, which is not available or affordable for many people. He added that the next goal for his is team is to “deploy this machine learning model and initiate an engagement strategy [where] the robot notices [if] the child is disengaged or losing focus. Then the robot steps in and gives some good feedback. Another thing is to expand to other behaviors besides engagement. There are a lot of behaviors children [with ASD] express [behaviors like frustration and excitement], and understanding their behavior and their emotional state is important. [Incorporating] things of that nature can really help create robots that are more personalized to each child and his or her unique needs.”



©2020 ICare4Autism. All rights reserved.

  • Facebook Social Icon
  • Instagram
  • YouTube Social  Icon
  • LinkedIn Social Icon
  • Twitter