Converting multi-modal task instructions to rule-based robot instructions

While frame-based representation of knowledge is a well known concept, its application to verbal rule instructions poses a number of problems. This paper describes how verbal instructions can be converted (mapped) into a frame-based representation, to form a base for reasoning and carrying out robot actions. Furthermore the paper introduces idea of corpus-based robotics, which supports the design of natural human-robot communication systems. An application and experimental results to the scenario of teaching a card-game are shown.