Personal peculiarity classification of flat finishing tool motion by using self-organizing maps for skill training

The paper proposes a unsupervised classification method for personal peculiarities in flat finishing motion, which are measured by three dimensional stylus attached to an iron file. The method classifies personal peculiarity variations in order to correct bad finishing motions effectively for skill training. In the case of flat finishing motion, the number of classes of personal peculiarities is unknown. A Self-Organizing Maps is used to classify such unknown number of classes of personal peculiarities. Experimental results of the classification with measured data of an expert and two learners show availability of the proposed method.