ABSTRACT
Factories experience problems such as accidents and product defects. Accidents cause injuries to workers, and sometimes cost them their lives in the worst case. Product defects drain customer confidence in vendors. They also increase production cost because of delays in production processes. One of the major causes for these problems is human errors. The introduction of factory automation (FA) technologies might be one way to solve these problems. However, the cost to introduce FA technologies is so huge that many working processes are still operated by human power. We should provide a method to prevent human error for processes when human power is dominant. Generally, human errors come from lack of worker concentration. This paper values body movement of workers while they show symptoms of human error. We think the worker have good rhythm and bad rhythm in monotonous work. The body movements of the worker detect rhythm. We have focused on the transition of the degree of similarity. The experiment makes workers to simulate repetitive works of tracing a circle shown on a tablet PC and displayed 100 times. This system finds a symptom of human error. We make sure that the transition of the degree of similarity has validity.
