Personality Synthesis Using Non-humanoid Cues

Abstract

Currently there exists literature and research done on the role of personality in robots. However the existing research involves exhibiting personality on some platform which contains human cues. On the other hand, there exists little research on attempts to synthesize and exhibit personality without the use of any humanoid cues. In our research, we explore this challenge. In concrete, we define four parameters that modulate the migration behavior (motion path and schedule) of an information-seeking agent in a virtual 2D environment where only moving triangles and static rectangles reside. The presented model will be applicable to any mobile social robots. We setup five different agent behaviors as video stimuli and asked subjects to evaluate the agent’s personality based on the Big Five personality traits model. The results suggest that people perceive different personalities from different parameter settings. We will discuss the future direction and issues based on the results.