A Study on Human Interactions With Robots Based on Their Appearance and Behaviour

When designing a natural-language interaction for a social robot, it is not enough to design the conversation itself: the success of a human-robot interaction can also be significantly affected by seemingly small factors such as a robot’s physical appearance and non-verbal behaviour. In this paper, we deploy an identical chatbot system onto two different robots, Furhat and Pepper, and compare users’ subjective responses to conversations with both robots to get a clear measure of the impact of robot appearance on a social robot when the interaction context is held constant. The results of the study were varied: Furhat was considered to display emotions better and to be more intelligent and trustworthy than Pepper, while both robots were seen as equally friendly. No significant differences were found in the likeability and comfort categories.