Back to listing

Teacher robots still have much to learn

Learning, Technology, University of Queensland

Date: 6 March 2020

Today’s babies grow up in a world of screens: they’re surrounded by televisions, computers and tablets, they watch adults glued to their phones. But the next generation of children may have a whole new layer of artificial media to contend with – social robots.

Robots in classrooms seem inevitable – so what are the best ways of integrating them into children’s learning environments?

That’s the question being answered by finishing PhD scholar Kristyn Sommer (University of Queensland) who researches children's capacity to learn from social robots. At CoEDLFest 2020, Kristyn brought her own first child Sienna to view some of mum’s recent work: a poster on ‘Preschooler’s Learning of Difficult Words: Do faces, robots or technology help?’ which she co-authored with Centre Chief Investigator Paola Escudero.

Kristyn’s highly interdisciplinary doctoral thesis investigated whether young children learn better from social robots than traditional screen media, such as iPads and TVs, which young children are notoriously bad at learning from. Her interest in studying social robots stemmed from the popular push to get kids to learn as well from iPads as they do from people, with social robots as one approach for doing so.

“We seem to really want an iPad to be something a child can learn from without the interaction of a parent,” Kristyn says. “So my plan was to introduce a social robot into that space, because it’s interactive, and it talks to them, and it can offer the social and emotional help you can’t get from a passive screen.”

However, Kristyn’s research concluded that a robot doesn’t really help much yet: the kids in Kristyn’s studies simply didn’t imitate the robots’ actions as well people. “A two-year-old learning from a robot gets about the same results as one-year-old learning from a human, for example in how well they can follow a set of actions to complete a task,” Kristyn explains. “Robots are pretty social, but they still don’t look, sound or move like a human, so a lot more research and development is needed to improve child-robot interaction.”

RUBI the Robot performs Kristyn’s imitation tasks

To complete her research, Kristyn was given significant access to RUBI, a kinetically sophisticated robot at the University of California, San Diego, kindly facilitated and co-funded by The Temporal Dynamics of Learning Center. Kristyn programmed RUBI for several studies while working with the Centre’s UQ team, which designs and builds various types of social robots, including Opie.

“We designed, tested and refined multiple Opies to be child friendly and super safe, with slow movements and back-drivable motors, unlike RUBI,” says Kristyn. “This way kids could Hi-5 and hang off Opie’s arms and really enjoy touching and playing with robots which traditionally prohibit kids from getting too close. “I took a field I knew well – Imitation – and first applied that to experiments with children and robots. Now with the help of people across the Centre I get to take on the more complex task of language learning with social robots.”

Kristyn also believes that instead of waiting for robots to become embedded in classrooms and then asking if they work, it’s better to test if they’re actually going to work, and how. “Rather than have governments spend $20,000 per classroom on a social robot, let’s see what it’s likely to do to the children and ask – is it worth it? And everything we learn can inform the design of future robots,” she adds.

So, what is the answer to the question posed in Kristyn and Paola’s poster? Do faces, robots or technology help? The study  looked at whether children can learn really challenging words (minimal pairs) from eBooks based on whether they’re embedded within a robot that is either very social or a little bit social, held by a person, or just on its own while the child uses an iPad.

Unexpectedly, the findings suggest that children learn the words best from the iPad alone.  “We suspect that the robots may actually be distracting the kids from the eBook because they are encouraging the kids to look away from the story and at their faces instead,” Kristyn suggests. “I’m looking forward to working with Paola to further investigate this unexpected finding over the next couple of years.” 

Kristyn Sommer with baby




  • Australian Government
  • The University of Queensland
  • Australian National University
  • The University of Melbourne
  • Western Sydney University