A Comparison of Two Approaches to Support Methods Time Measurement in an Automotive Factory

In a manufacturing factory, eliminating seconds off a task that is repeated often can reduce the overall manufacturing time and improve the profitability of each unit produced. It is advantageous to understand the average amount of time required to complete each task. In order to scientifically determine the desired time required to complete a task, the individual subtasks also can be timed. This can be made more efficient by automating the process of recording the basic physical motions of factory workers that are involved in a task. Our paper shows how it is possible to use a machine learning based approach to classify the basic motions. This paper describes two approaches that we implemented in order to automate the process of motion classification. We contrast the two approaches and analyze the tradeoffs between each approach. The context for the application of our project is Mercedes-Benz US International, a large automotive manufacturing facility in the Southeastern United States. Additionally, we discuss the limitations of the two approaches and future work that can address these issues.