Knowledge Acquisition with Selective Active Learning for Human-Robot Interaction

摘要

This paper describes an architecture for robots interacting with non-expert humans to incrementally acquire domain knowledge. Contextual information is used to generate candidate questions that are ranked using measures of information gain, ambiguity, and human confusion, with the objective of maximizing the potential utility of the response. We report results of preliminary experiments evaluating the<br/>architecture in a simulated environment.