Can robots be good therapists?

Using therapeutic robots to support people’s needs is a rising field of research in healthcare. Designing them for best results, however, is more than an engineering problem according to Paul Bucci, a Computer Science PhD student at UBC.

A variety of robots are currently available for assistive care. PARO, a white baby seal robot from Japan, is perhaps the most well-known. It has been successfully utilized for dementia care in multiple countries since 2003.

Paul Bucci, UBC computer science PhD student, with objects used to investigate people’s emotional and physical responses | Photo courtesy of Paul Bucci

A number of studies on therapeutic robots have been done to evaluate their effects. After working closely with sick children at the Canuck Place Children’s Hospice in 2019, Bucci realised robot designers can learn from therapists.

“With my work in particular, we were using the robot that I had developed in my Master’s degree and we were looking at how we might continue to create touch-based robotic interactions at the Children’s Hospice,” Bucci says. He and his team developed CuddleBits, furry robots that can be handheld and are more mobile and robust than larger robots.

“It’s a very complex, complex field. A lot of our findings were more engineering questions of how do you actually do the work of creating the interactive programs and the physical structures themselves,” he explains.

A human-centered approach

But beyond that, he suggests that researchers and designers need to think about emotions in the ways that therapists do – that they aren’t inherent or experienced the same way by everyone.

An example of a robot designed for therapy. | Photo courtesy of Paul Bucci

He adds that the current studies tend to lock down a very precise mathematical description of what goes on in terms of specifying what the robot and the human are behaving and feeling, but he doesn’t think that is the right approach.

“I think that the models that we were using were not complex or accurate enough in terms of human emotions. What I’m suggesting is a human-focused approach – to work more with the interaction itself and less with an engineering-focused approach,” he explains.

Bucci says designing interactions doesn’t take very advanced artificial intelligence. It is more about the narrative.

“We did a study where we had the exact same robotic motions given to 10 to 20 different people. What seemed to change their emotional reactions was actually the narrative that they placed on the robot. Literally the same motions didn’t really matter because it was all about the story that they brought,” says Bucci.

Based on his findings, past experiences should be included when analyzing emotional interactions. For example, someone who was bitten by a dog in the past would not feel as comfortable with a robotic dog as others would.

“If we were to change our thinking about emotions and how to measure them, we wouldn’t waste time designing and running studies that don’t really measure their goal, and ultimately, we could build more emotionally impactful robots,” he says.

Bucci thinks the future of designing therapeutic robots really depends on the therapists. Designers and engineers need to work with them to understand the therapeutic technique and how it would look to incorporate interactions with robots into the practice.

“The funny thing is that people tend to attribute agency and emotion to anything, in which case it means we need to design the robots to be intelligent as much as very reactive,” he says.

He concludes that technology can do amazing things, but it doesn’t replace people. So, when designing technology, it is always better to keep in mind people’s need for interaction with other people and the emotional implications of such interactions.

For Bucci in the end the robots are only tools to enable therapy; it is not the robot that makes for effective therapy but the human agent.

For more information visit: