Can robots help prevent anxiety attacks in children?

Morten Roed Frederiksen from Computer Science at ITU has received 1.6 million kroner from Independent Research Fund Denmark for at research project that aims to make robots better at understanding human emotion. The goal is to create technology that may help children with anxiety.

Computer Science DepartmentResearchalgorithmsartificial intelligencegrantsrobotsMorten Roed Frederiksen

Written April 26, 2022 9:42 AM by Theis Duelund Jensen

When people interact, we automatically adjust to our surroundings and to each other. If you are having dinner with a friend at a packed restaurant, you automatically raise your voice, so your friend can hear you. If your friend is in a bad mood, you adjust your tone of voice according to the emotional response triggered. That is how we humans communicate with each other – by adapting to the context we are part of.

Robots and other types of artificial intelligences do not possess the same finely tuned situational awareness as we do. Unlike people, robots do not have an innate ability to filter unnecessary information. Thus, researchers are preoccupied with constructing algorithms and technologies that may allow artificial intelligences to filter information more efficiently – especially when it comes to adapting behaviour in relation to people.

In his work, postdoc and affiliate researcher of Robotics, Evolution, and Art and AIR at ITU, Morten Roed Frederiksen, attempts to increase artificial intelligences’ abilities when it comes to adapting to humans. His project Deep neural network based physical response prediction using correlated patterns in dynamic sensor input for adaptive robots has recently been awarded 1.6 million kroner from Independent Research Fund Denmark ear-marked for research at University of Southern California in the USA. Here, Morten Roed Frederiksen plans to work on artificial intelligence technology that may help children who suffer from anxiety attacks.

Robotic empathy
“Today, children suffering from anxiety are taught to use so-called focus objects to distract their attention from an oncoming anxiety attack. That means the onus is on the child to deal with the anxiety attack. I have conducted a lot of research in the field of social robots, and to me there seems to be a lot of potential in using robots and artificial intelligence to help children deal with anxiety attacks,” says Morten Roed Frederiksen.

In California, the researcher will be working with leading expert on social robots, Professor Maja Mataric, who has previously developed so-called socially assistive robots that help children with autism navigate social situations. Morten Roed Frederiksen’s goal is to develop a pocket-size robot that reacts to early signs of an oncoming anxiety attack.

“We want to find out if we can make a robot capable of understanding if a particular context is unpleasant to a child. That is the first step towards understanding social context in general. If we can enable a neural network with that ability, it will not only be a big contribution to robot science – it may also help a lot of children who suffer from anxiety,” says Morten Roed Frederiksen.

Morten Roed Frederiksen will be training the empathetic robots on an audio dataset. By finding common audial denominators in various scenarios, the researcher plans to identify potential anxiety attack triggers in a group of test subjects.

“Prior research has shown that artificial intelligences may be trained to perceive emotionally charged words. The road to an artificial intelligence that can predict an emotional response is however long. For now, I am excited to get started on the project, and I am happy that IRFD sees the potential in my idea. The grant is huge recognition of my work,” says Morten Roed Frederiksen.

Further information