Title: Conflict Detection and Resolution in Table Top Scenarios for Human-Robot Interaction

Abstract: As in any interaction process, misunderstandings, ambiguity, and failures to correctly understand the interaction partner are bound to happen in human-robot interaction. We term these failures 'conflicts' and are interested in both conflict detection and conflict resolution. In that, we focus on the robot's perspective. For the robot, conflicts may occur because of errors in its perceptual processes or because of ambiguity stemming from human input. This poster presents a brief system overview, and details Here, we briefly outline the project's motivation and setting, introduce the general processing framework, and then present two kinds of conflicts in some more detail: 1) a failure to identify a relevant object at all; 2) ambiguity emerging from multiple matches in scene perception.