ASSESSING LEARNING OF MANUAL ASSEMBLY PROCESSES USING WRIST-WORN INERTIAL MEASUREMENT UNITS

The objective of the research presented here is to study the use of Inertial Measurement Units to assess how consistently a person performs a manual assembly process. In conjunction with this objective, the research also explores the use of two novel metrics to plot learning curves - cycle-to-cycle consistency, and consistency with average. These new metrics allow for insight into the progress a person is making while learning how to perform an assembly process. This is a deeper level of insight than the prevalent approach of using cycle times as a metric to plot learning curves. Experiments are conducted where three co-authors performed the assembly of two products fifteen times each. They wore IMUs on both wrists which allowed the difference in consistency between dominant and non- dominant hand to be analyzed. Pearson Correlation Coefficient between the consistency-based learning curves and the cycle time-based learning curves were computed. It was found that these learning curves are weakly correlated, thus indicating the need for both to be considered individually while evaluating whether a trainee is ready for a live assembly line or not. This research needs to be strengthened by performing experimental tests where the participants are not related to the study, by collecting more cycles of data for a larger variety of products, and by including a metric of correctness. This is needed to ensure that people perform the correct process consistently.