Development of Training System for Appearance Inspection Using Motion Capture and Large Size Display

To assure appearance quality of industrial products, appearance inspection by human vision is implemented in many manufacturing industries. In appearance inspection by human vision, small visual defects such as dirt, scratches, surface dents, and unevenness of the coating color are inspected. In actual defects in appearance inspection processes, the visibility differs depending on the positional relationship between the inspector and the defect. Furthermore, visibility varies depending on the type and characteristic of defects. Therefore, to comprehensively inspect these defects, it is required to change the positional relationship between the inspector and defects by appropriate body movements of the inspector. In particular, in case of product size is relatively large such as a body, glass of an automobile, there is strongly required the body movement of the inspector to realize positional relationship in which defects can be detected. However, no studies have been conducted on appropriate body movements based on the defect positional relationship between the inspector and defects, in actual appearance inspection processes, work training is only carried out by on the job training based on experience of skilled inspectors. Therefore, this study focused on appearance inspection that required appropriate body movements of inspectors during the inspection, and also developed a work training system for appearance inspection using motion capture and large size display. Moreover, to show how to utilize the developed training system, the experiment using the general work procedure manual and the work procedure manual that teaches detailed body movements according to size of products and height of an inspector are designed as experimental factors, and their effects on defect detection are evaluated with 12 subjects. As a result, it is quantitatively obtained that the latter work procedure manual has the effect of improving the defect detection rate about 5.9%. In particular, it is also obtained to be effective in case of detecting small luminance contrast and detectable range of defects. From the above, it is considered that the work training methodology including body movements is effective, and it is shown that the developed training system may contribute to work design and work training in the actual appearance inspection processes.