How to talk to robots: Evidence from user studies on human-robot communication

Abstract

Talking to robots is an upcoming research field where one of the biggest challenges are misunderstandings and problematic situations: Dialogues are error-prone and errors and misunderstandings often result in error spirals from which the user can hardly escape.Therefore, mechanisms for error avoidance and error recovery are essential.By means of a data-driven analysis, we evaluated the reasons for errors within different testing conditions in humanrobot communication and classified all the errors according to their reasons.For the main types of errors, we implemented mechanisms to avoid them.In addition, we developed an error correction detection module which helps the user to correct problems.Therefore, we are developing a new generation strategy which includes detecting problematic situations, helping the user and avoiding giving the same information to the user several times.Furthermore, we evaluate the influence of the user strategy on the communicative success and on the occurrence of errors within human-robot communication.In this way, we can increase the user satisfaction and have more successful dialogues within human-robot communication.