Filling Knowledge Gaps in Human-Robot Interaction Using Rewritten Knowledge of Common Verbs

In this paper, we present an approach to representing a core part of the knowledge consists of semantic information of common verbs from semantic dictionaries. We provide a metalanguage as the representation framework for the rewritten knowledge of common verbs and their corresponding user tasks. The metalanguage is interpreted based on transition systems, which can be realized on various formalizations such as situation calculus, action languages, and answer set planning. We realize the approach based on answer set planning. Moreover, we provide empirical evidence showing that HRI may significantly benefit from the rewritten knowledge and remarkable performance improvement compared to previous work.