Author/Editor		Dillmann, R; Kaiser, M; Ude, A
Title		Acquisition of elementary robot skills from human demonstration
Type		članek
Source		In: Proceedings of the 3rd international symposium on Intelligent robotic systems '95; 1995 Jul 10-14; Pisa,
Publication year		1995
Volume		str. 185-92
Language		eng
Abstract		Similarly to human voluntary movements, a robot's elementary skills can coarsely be divided into two main categories: rapid transfer movements that are executed in an open loop with no opportunity for correction and slower controlled movements guided by sensorial feedback (6). In this paper we consider two particular problems which arise when acguiring open-lood and closed-loop elementary skills: the segmentation of the trajectory into time intervals corresponding to different classes of the operator's motion and the preprocessing of the possibly bad examples that were recorded during a demonstration in order to make them suitable for skill learning. The proposed methods are based on the fuzzy set theory, statistics, and Shannon's information theory.
Descriptors		MOVEMENT ROBOTICS MODELS, THEORETICAL