Assembly skill transfer system for cell production

Feng Duan,Qi Gao,Tamio Arai

DOI: https://doi.org/10.1109/ROBIO.2010.5723336

2010-01-01

Abstract:This study aims to develop a skill transfer system to extract assembly skills from skilled operators and transfer them to novice operators. To achieve this aim, the fundamental problem is how to extract assembly skills from the skilled operators in a systematic way. Considering the characteristics of assembly tasks in cell production, a human assembly skill model was proposed, which consists of cognition skill and motor skill. Although many assembly skills exist, only the representable ones (can be represented by human understandable natural language) can be extracted and transferred to novice operators. Even novice operators may know some assembly skills based on their common senses. In this case, only extracting and transferring the assembly skills that skilled operators emphasize on but novice operators do not emphasize on may be an effective way. Therefore, the degree of emphasis is proposed to obtain how much the operator emphasizes on a skill. When extracting the skills, a group of skilled operators and novice operators were required to execute a cable harness assembly task, and their degrees of emphasis and assembly time were scored. After calculating the correlation between the degrees of emphasis and assembly time, the correlation coefficients of the skills can be obtained. Based on the proposed method, the assembly skills with the maximum absolute correlation coefficients were transferred to novice operators. The results show that after transferring the extracted assembly skills, the novice operators' assembly performance was improved greatly. On the basis of the skill extraction method, a multi-modal based assembly skill transfer system (MASTER) was developed. Evaluation results show that working under this skill transfer system, novice operators' assembly performance can be improved greatly.