Vanderbilt graduate student Pramila Rani is hooked up to biosensors that detect and analyze different emotional states.

Professors Sarkar (standing) and Smith tested their findings on a small, mobile robot.

"Psychological research shows that a lot of our communications, human to human, are implicit," says Nilanjan Sarkar, an assistant professor in mechanical engineering. "Sooner or later, robots will be everywhere. As they become increasingly common, they will need to interact with humans in a more natural fashion."

Researchers are now collecting baseline information about subjects and analyzing it to identify specific responses associated with different mental states. Initial tests concentrated on detecting high and low anxiety levels. Researchers hooked subjects to heart-rate monitors and had them play video games with varying levels of difficulty to induce stress. They then used advanced signal-processing techniques, such as wavelet analysis and fuzzy logic, to analyze the data.

What they found wasn't a surprise: With stress, heart rates go up. Researchers combined the heart-rate data with measures of skin conductance and facial-muscle activity to create a set of rules that would let a robot respond to information about a person's emotional state.

They then tried the signals out on a small, mobile robot. Researchers input physiological data of a person with high anxiety levels, then instructed the robot to move to a specified location and ask, "I sense that you are anxious. Is there anything I can do to help?"

One area of continued study is on finding a way to discriminate between high anxiety levels and engagement, say researchers.