On The Emotional Transparency of a Non-Humanoid Social Robot

Non-anthropomorphic robots have issues in conveying internal state during a Human-Robot Interaction (HRI). A possible approach is to let robots communicate their states or intentions through emotions. However, the robot’s emotional responses are not always clearly identified by people, and it is also difficult to identify which and how many cues are most relevant in affecting people’s ability of recognition of robots’ emotions during the ongoing interaction. We involved 102 participants in an online questionnaire-based study where they rated the robot’s behaviours, designed in terms of colours, movements and sounds, according to the perceived emotions in order to identify the cues to be used for making robots more legible. The results suggest that emotional transparency can benefit from multimodal interaction. Our results underline that single modes can be capable of conveying effectively the desired emotion, and little benefit is obtained by the use of additional modes that may be not necessarily noticed by the users.