Understanding of teaching information by autonomous mobile robot-acquisition of skill rules

This paper discusses teaching information understanding by an autonomous mobile robot. The authors have studied a system that has a function of input variable selection and rule extraction from data for the grounding of teaching information. The performance of acquired rules was often degraded in the new environment, and refinement of the rules became necessary. This paper discusses the refinement of the acquired rules. We call the refined rules "skill rules". Through interactions with the new environment assisted by a human via the environment, the acquired rules are to be refined. Simulations are done to demonstrate the skill rule acquisition.