Closing the Feedback Loop: The Relationship Between Input and Output Modalities in Human-Robot Interactions

Previous studies suggested that communication modalities used for human control and robot feedback influence human-robot interactions. However, they generally tended to focus on one part of the communication, ignoring the relationship between control and feedback modalities. We aim to understand whether the relationship between a user's control modality and a robot's feedback modality influences the quality of the interaction and if so, find the most compatible pairings. In a laboratory Wizard-of-Oz experiment, participants were asked to guide a robot through a maze by using either hand gestures or vocal commands. The robot provided vocal or motion feedback to the users across the experimental conditions forming different combinations of control-feedback modalities. We found that the combinations of control-feedback modalities affected the quality of human-robot interaction (subjective experience and efficiency) in different ways. Participants showed less worry and were slower when they communicated with the robot by voice and received vocal feedback, compared to gestural control and receiving vocal feedback. In addition, they felt more distress and were faster when they communicated with the robot by gestures and received motion feedback compared to vocal control and motion feedback. We also found that providing feedback improves the quality of human-robot interaction. In this paper we detail the procedure and results of this experiment.