Researchers from the University of Southern California have presented AI-empowered socially assistive robotic solution to help children with autism spectrum disorders.

In the article “Modeling engagement in long-term, in-home socially assistive robot interventions for children with autism spectrum disorders” published in journal Science Robotics on February 26, Interaction Lab researchers have presented machine-learning algorithms that are capable of tuning to the particular child needs.

While usually human-robotic interactions are “still limited in their ability to autonomously perceive, interpret, and naturally respond to behavioral cues from atypical users in everyday contexts”, reads the research paper, AI algorithms that use audio and video data inputs, for example from dialogues or eye contact, can help robot predict the child engagement, and therefore personalise the interaction.

During the study, when researchers have been collecting data from robots for a month-long period, the AI-based robot accuracy rate of engagement predictions was 90%, reported the scientists.

Since therapy sessions that are offered for children with autism spectrum disorders could be expensive and are generally limited to few hours per week, new robotic solutions can be of a great assistance, given that AI-empowered system can hold the child’s attention to the therapy exercises.

Additionally, according to co-author Maya Mataric, “kids need to learn in a social setting”, and since kids with autism disorders do not usually have enough practice with social learning, robotic solutions are very important.

In future, researchers plan to minimise the data inputs, necessary for the AI algorithm to learn, with a goal to protect children privacy. Moreover, the team is working on developing more affordable personalised therapeutic robotic companions.

Follow us on Facebook, Twitter, Telegram, or Youtube!