Pushing the boundaries of social robotics
What is it about a person that makes others open up to them? You know the type – that person in the office that most people feel comfortable with.
What if that person could be a robot? What would that robot require in its programming to make others feel safe? Would it have a voice? And if it did have a voice, what would that voice sound like?
These are questions being worked on by robotic engineers Scott Heath and Gautier Durantin. Both are postdoctoral researchers at the ARC’s Centre of Excellence for the Dynamics of Language (CoEDL). Their research team considers timing to be a key to social interaction.
“If we are making a robot that can communicate with a voice then timing is critical,” says Scott. “We draw on a lot of research from our colleagues at CoEDL and CoEDL affiliates and this research is showing us that gaps between speakers conversing can be as small as 200 milliseconds and sometimes there are no gaps; people are often responding to a sentence before the previous speaker has finished.”
Scott and Gautier say creating technology that can even partly mimic friendly human interaction is challenging. Even one-way engagement is reliant on sophisticated timing says Gautier. “Story telling for instance is a very human ability and timing in story telling is just as important as in conversation.”
Both researchers are working as part of a CoEDL and University of Queensland team experimenting with a robot called “Opie”. Opie has already been at schools, where children are fascinated by his look and ability to tell illustrated stories. “Most users of technology today adjust to the technology but our group’s aim is to adapt the technology to the user,” says Gautier. So with this in mind, Opie is being tested in classrooms. “We are keen to continue to adapt Opie to the children,” says Gautier. “Interactivity, adaptability and accurate timing are all aspects of robotics that are needed if we want to build better interactions.”
“Social robotics is still in its infancy,” says Scott. “Robots currently available off the shelf don’t have the ability to engage at a pace fast enough for most children. So while children are fascinated to start with, they get bored quickly.”
Another issue is how to program a robot to better respond to the meaning of a sentence. “What I’m particularly interested in is how robots understand and respond to language,” says Scott. Humans respond to language with multiple senses so a robot needs multiple senses also. “For instance a camera can assist a robot to understand colour.”
Working within CoEDL allows the robotic researchers to collaborate with linguists and psychologists, and to combine expertise across the language spectrum.
Opie, and the researchers developing him, will continue to test the boundaries of what is possible. Scott says head of his research group, Professor Janet Wiles, has a great metaphor for how the team thinks about technology. “She points out that when the Wright brothers invented a flying machine, they were inspired by birds but didn’t directly copy the flapping motion of the birds,” he says. In the same way the researchers know that the robots they develop are unlikely to be able to replicate all human traits but neither will they have to in order to be successful.
Image above: Dr Scott Heath and Dr Gautier Durantin discussing problems associated with building early prototypes of social robots.
Main image: Some of the Opal Team who work on developing Opie as a social robot. Leaning in is Kristyn Hensby with Janet Wiles (back left) Jon Taufatofua, Gautier Durantin, Marie Boden and Scott Heath.