Managing Scripts for Interactive Service Robots by Using Action-Relation Rules

An approach in which manually written task procedures are used is proposed for generating natural and comprehensive task plans. However, semantic inconsistencies may occur in manually programmed procedures. In this paper, we propose a script management system on the basis of action-relation rules that have been generated from patterns mined by the Prefixspan algorithm. Action-relation rules generated in an interactive manner act as a guide for eliminating or modifying task plans. This paper presents the principle and user interface of the management system and the result of implementing the proposed method.