A study on skill acquisition from teaching information by mobile robot

Abstract

This paper studies teaching information understanding by mobile robot from a point of view of skill acquisition. This paper considers skill for a robot as "the learnable part which mostly affects the robot performance". The skill in this sense is in the refined parts of preprocessing of input variables, control rules, output variables, and also in the sub-evaluation functions to be incorporated into the robot. The process of teaching information understanding is thus viewed as a process of acquiring the above skills. This paper studies acquisition of skills for task execution by using actual mobile robot. This paper focuses on the skill acquisition from expert's assistance given in the form of patterns. Skill acquisition is to acquire refined rules. This refinement of rules is achieved by re-dividing the input space. The skills are explicitly acquired as differences between the rules before/after the assistance by the expert. Experiments are done to demonstrate the skill acquisition using a mobile robot.