Evaluating Customers’ Engagement Preferences for Multi-party Interaction with a Robot Bartender

In this paper, we present a two parts study where we investigate whether socially relevant interactions depend on a bartender robot’s ability to maintain engaged interactions with two users at a time, by alternating its attention between the two customers, and by capturing their attention by personalising the discussion topics to the users’ preferences. We recruited 20 participants in the first study, and 6 participants in the second study. We observed that participants were more comfortable and enjoyed interacting with the robot when the robot personalised interaction and alternated the service between the two users, compared to when the robot did not personalise the talk and served the users according to their arrival time. However, we observed that in both conditions, the participants did not feel excluded while waiting for the other to finish their interaction with the robot. Finally, personalised interaction topics increased participants’ interest in the conversation with the robot. These results guide us to design and deploy a robot that adapts its behaviours to have rich and multi-party interactions in real settings.