Concept of Fuzzy Atmosfield for Representing Communication Atmosphere and its Application to Humans-Robots Interaction

Abstract

The concept of Fuzzy Atmosfield (FA) is proposed to represent a communication atmosphere in society consisting of multiple individuals such as humans and/or robots, where the FA is characterized by 3D fuzzy cubic space with “Friendly-Hostile” (FH), “Lively-Calm” (LC), and “Casual-Formal” (CF) axes, and each state of the FA is visualized using shape-colorlength graphics. It is targeted to be a tool for identifying an atmosphere using quantitative analysis and graphical representation. By a humans-robots interaction experiment in which the FA is used to represent the real-time atmosphere created by four humans and five eye robots in a home party scenario, it shows that Pearson’s correlation coefficient values of 0.92, 0.86, and 0.72 for the FH, LC, and CF axes, respectively, indicate the correspondence between the proposed FA and results of questionnaires, and that subjective estimation of graphical representation of the FA achieves 84% accuracy for shape, 76% for color, and 58% for length. The FA is being extended to the representation of complex atmosphere generated by humans, robots, and background music, and part of results is also shown.