As robots become more prevalent and human-like, roboticists face the daunting task of bridging the uncanny valley. Since the face is typically the platform by which we communicate and express ourselves, the robots we create better have faces we feel comfortable engaging with. So, what should these faces look like?
Akanksha Prakash, a graduate student at Georgia Tech’s School of Psychology, had that same question, so she conducted a study to find an answer. Participants of varying ages were shown a series of three images — a human face, a robotic face or a combination of the two — and asked which one they felt most comfortable interacting with.
Most college-aged participants preferred a robotic face, but they were generally open to others options. On the other hand, almost 60 percent of older participants chose a more human face, and only 6 percent of them selected a mixed human-robot face. However, preferences in both age groups fluctuated when participants were told the robot would be helping with chores, personal care, decision making and social interaction.
“We found that participants, both younger and older, will assign emotional traits to a robot based on its face, which will determine what they are most comfortable interacting with,” Prakash said in press release. “As a result, preferences for robotic appearance varied across tasks.”
Chores did not elicit many strong preferences, but personal care, such as bathing, did. Both age groups were divisive on the subject, with some associating a more human face with trustworthy nursing traits. Due to the private nature of the task, others didn’t want anything human-like bathing them and chose more robotic features.
For decision-making tasks, such as investment advice, younger participants chose a human-robot face, while older participants preferred a more human face. “Those who selected a mixed face perceived the robot as more intelligent, smarter or wiser than one with a ‘cute’ robotic face. Perceived intelligence in appearance was an important assessment criterion for receiving assistance with decision-making tasks,” said Prakash.
For social interaction tasks, both ages preferred the human face. Prakash says her research could one day help developers and engineers design robots with faces based off of designated tasks. Prakash wants to expand the study to include other age groups and participants with a more diverse educational background.
Credit: Georgia Institute of Technology