Applying the concept of "Moe" to designing robot's appearances

Abstract

One of the problems with human-robot interaction is how to design robot's appearances. It is said that robot's appearances can affect the emotions of people who communicate with the robot. The "uncanny valley" proposed by Mori is a famous phenomenon with regard to robot's appearances. It is necessary to avoid the "uncanny valley" when designing a robot. We propose adopting the concept of "moe" (a Japanese slang word expressing feelings such as goodwill, love, and excitement to the characters in comics or animated cartoons) as a design element to design a robot that most people will be acceptable. "Moe" is an ambiguous concept, and utilizing it for designing robot's appearances is not a simple task. Accordingly, a 3D design system for investigating and clarifying the meaning of "moe" was built. The system is based on an interactive genetic algorithm (IGA). The system was used in experiments to compare "moe" with the English words "prettiness" and "coolness". In particular, the possibility that there is a difference between appearances of robots designed on the basis of whether people understand "moe" or not was investigated. In this paper, it appears that using "moe" and using "prettiness" as concepts for robot's appearances are equivalent under the condition that the robot's appearances was primitive.