Evaluation of Social Robots with Social Signals in Public Spaces

Evaluating the users’ subjective impression of social robots with questionnaires in field trials is time-consuming, and it is impossible to analyze every interaction as users cannot always participate for various reasons. In social robotics, social signals (e.g., gestures, facial expressions, body language) can be used to control and adapt the robot’s behavior to improve interaction. Social signals should allow a statement to be made about how the user feels in a situation or what attitude or opinion they have about something. This paper focuses on finding out to what extent social signals can contribute to an automatic evaluation of the subjective user satisfaction without the need for user questionnaires and solely through social signals. For this purpose, it is first relevant whether social signals occur and if they can be used in the present use case. For this purpose, a field test was carried out in the entrance area of a city administration, in which the social robot Furhat was used as a point for service and information. Four interactions were recorded on video with the consent of the users, and a post-questionnaire on satisfaction was collected for each case. A qualitative video analysis was used to examine the interactions concerning potential social signals displayed by the users. The results suggest that the direction of view and the head orientation are the body parts that are mostly moved in the interaction. All users interacted with the system with relatively little movement of the rest of the body. The results suggest that it is not recommended to focus only on social signals alone.