I’ve written a few times about the growing role robots are playing in a number of healthcare scenarios, and in particular how humans tend to respond to the presence of robots in such sensitive settings.

A team from the University of Bristol have further examined the use of robots in healthcare by testing their use in the rehabilitation of people suffering from social disorders.

All interaction is good interaction

The work, which was documented in a recently published paper, revolves around the theory of similarity, which suggests that we find it easier to bond with those who are like us, whether that’s in looks or behaviors.

The study involved patients playing a version of the mirror game, which sees participants attempting to copy the motions of the other player, whilst at the same time playing with some colored balls on a string.  In this instance, the participants were paired up with either a robot or a digital avatar.

The avatar was initially programmed to look and behave like the patient to try and facilitate an attachment with them.  Over time however, the avatar would become less like them, with the intention being to promote the social rehabilitation of the patient.

When the results were analyzed, the researchers found that indeed there were connections forged between players who noticed similarities in behavior, even if those players weren’t in any way real.  The authors believe this could be a crucial finding for the successful rehabilitation of patients with social disorders as the artificial peers can be programmed to have the same traits as the patient.

“It is very challenging to build an avatar that is intelligent enough to synchronise its motion with a human player, but our initial results are very exciting,” the authors say.

The approach is interesting, not least because the system was smart enough to successfully mimic the motions of the human player.  It achieved this via dynamical systems principles and a feedback control approach.

The next stage is to build upon these initial results and test the method out in more complex, multiple user environments to allow groups to interact together.