By COLLEEN CREAMER
For The Nashville City Paper
It may be more little brother than big brother now, but Vanderbilt researchers Nilanjan Sarkar and Craig Smith are in the early stages of creating a robot that can sense human psychological states. It would be the first step in developing a kind of “robot Friday.”
The project will unfold in two steps, the first of which will be to detect human psychological states by analyzing the output of a variety of physiological sensors more complicated than simple body language. The second will be to process the states in real time and convert them into a form the robot can process.
Sarkar said psychological research now shows that the expression of psychological states is more individual than previously believed.
“The way we are trying to impart psychological states is through physiological signals. There is no unique signal for everybody,” Sarkar said. “But even though the signals are somewhat different from person to person there are similarities so you don’t have to learn from scratch for each person.”
Other factors such as culture, age and gender further make understanding and programming emotional states more difficult, he said.
“Change of heart rate is not the same for all people,” Sarkar said. “It’s very contextual, and some people show more activity and heart rate and less activity on the facial expression or vice versa.” Sarkar said the first experiments focused on high and low anxiety levels using a heart rate monitor, but the techniques proved slow and costly and not appropriate for a person moving and working.
One advantage, he added, was recent advances in minute sensor technology. By collecting data about stress, the two researchers developed a series of rules. They then used the rules to program a small mobile robot to randomly move throughout a room, sense unease and then say, “I sense that you are anxious. Is there anything I can do to help?”
Sarkar said the team was looking at a variety of uses. One, he said, could be for orthopedic rehabilitation. He said they were studying how to differentiate a person in distress from one that was merely concentrating. Understanding and implementing those nuances, Sarkar said, would then allow them to develop programs for people who needed to really focus and be challenged.
“There are a couple of different directions we’d like to go,” Sarkar said. “One is to have more psychological data not only limited to anxiety or stress. We want to understand engagement, the way in which a person is paying attention to the task and whether the person is getting frustrated or not.” Sarkar said the robot could then organize the person’s program in a way that maximized his or her potential without pushing the person’s boundaries.
–From The City Paper
