INVESTIGATING PERCEPTUAL AND BIOLOGICAL FEEDBACKS IN HUMAN ROBOT INTERACTION

The Human Robot Interaction (HRI) is a new discipline that has attracted more attention in the last years due to the increasing presence of robots in people’s everyday life. It is a field of study dedicated to understanding, designing, and evaluating robotic systems for use by or with humans. Interaction, by definition, requires communication between robots and humans. A social robot is an autonomous or semi-autonomous robot capable of interacting and communicating with humans or other autonomous physical agents following social behaviours and rules related to its specific role. In order to interact effectively with the human being, a robot must be able to decode the complex system of human clues during an interaction. This ability is innate in man, as our brains are accustomed to decoding the behavior and attitude of those in front of us through not only actions or words but also through spontaneous signals that show our partners’ intentions. The goal of my research is to to investigate human’s state, during the interaction with a robot, by the use of human’s biological and perceptual feedbacks. In particular, I want to provide an answer to the following research question: is it possible to model human’s response, using features that represent human behaviour, during the interaction with a robot? To investigate my research question, I considered different HRI paradigms corresponding to the humans’ perception of a robot. For each paradigm I derived some features that I explored in a dedicated experimental scenario. The approach used in my investigation has been multidisciplinary as it required to keep into account concepts and techniques borrowed from psychology, biology, cognitive science and medicine. I considered three state of the art paradigms for human-robot interaction: robot as Avatar, robot as Teammate and robot as Social Mediator. These paradigms are distinguished by the mental attribution that the person gives to the robot and by the role of the robot within the interaction. In the robot as Avatar the robot is seen as a projection of the person in charge of it, to communicate or interact with others. In robot as Teammate paradigm, humans and humanoid robot share the workspace and the objects to complete a task. in robot as Social-Mediator paradigm, the robot is used for conveying emotion or to support cooperation or learning. The user’s response in the interaction has been modelled deriving different features. In particular, I considered the following features: Biological feedback extracted from humans’s neurological activity. Acceptance assessed by questionaries. Trust derived from human’s neurological activity. Honest signals derived from human’s mimicry. Emotional Response aquired from an audience during a concert. To realize the experimental set to test each feature, I used four different robots. The Telenoid robot, which is a humanoid minimalistic robot used as avatar and social-mediator. The Geminoid robot, which is a human full scale humanoid robot. The Nao robot, which is a small size humanoid robot and the Kuka KRC 210 which is an industrial robot. The experimental results of the experiments carried out to validate the research showed the validity of the considered features. In fact, from the results obtained for each study it is possible to derive features to measure humans’ feedback. The further evolution of this research will be to equip a robot with a system based on these features to make the robot able to decode human beings mental and perceptual state and to react accordingly to their states and emotions.