| Despite tremendous advancements in the field of robotics, truly interactive robots that can communicate naturally and intuitively with humans have not yet been developed. As robots and people begin to co-exist and cooperatively share a variety of tasks, "natural" human-robot interaction (HRI) that resembles human-human interaction is becoming increasingly important. During human interactions, explicit communication transmits overt messages, while implicit communication transmits hidden messages about the speakers themselves (for instance their intentions, attitude, likes and dislikes). Sensitivity to the other party's emotions or affective states is one of the key attributes associated with the second, implicit channel. Even though emotions have been traditionally considered detrimental to rational functioning and maladaptive in nature, it is now being acknowledged that they play a pivotal role in "rational decision-making, perception, learning, and a variety of other cognitive functions". Therefore, endowing robots with a degree of emotional intelligence is expected to permit more meaningful and natural HRI.; This dissertation presents a psychophysiology-based affect-sensitive human-robot interaction framework. In this framework a robot interacts with a human based on his/her probable affective state, which is inferred from the human's physiological signals. The robot adapts its behavior in response to the human's affective state. Peripheral physiological signals such as electrocardiogram, electromyogram, photoplethysmogram, impedance cardiogram, and phonocardiogram were measured through wearable biofeedback sensors and new control architecture was designed to realize such HRI. Experiments were designed and implemented to verify anxiety feedback based HRI framework. The results showed that such a framework allowed better interaction between humans and robots by reducing humans' anxiety while improving their performance and overall satisfaction. |
