Referring Expression Generation Under Uncertainty in Integrated Robot Architectures

Tom Williams and Matthias Scheutz

Robotics: Science and Systems Workshop on Human-Centered Robotics: Interaction, Physiological Integration and Autonomy

To engage in language-based social interaction with humans, robots must be able to refer to other agents, objects, and locations, also known as referring expression generation (REG). Unfortunately, classic REG algorithms like the Incremental Algorithm assume that information is certain, centralized, and easily accessible: assumptions which commonly break down in realistic human-robot interaction scenarios. In this work, we show how the incremental algorithm can be extended to produce DIST-PIA: a domain-independent algorithm for REG under uncertain, heterogeneous, and distributed knowledge.