Analysis of manual manufacturing processes using motion sensing technologies

Abstract

To evaluate motion sensing technologies capable of collecting data that supports the analysis of workers in manufacturing environments, we developed a procedure to categorize manufacturing processes and many motion sensing technologies. Processes were decomposed into: resolution of movements, amount of noise, and detection difficulty; sensors were decomposed into: sensitivity, noise-cancellation capability, and detection capability. By analyzing each sensor alternative and checking if it provided the required functionality and level of performance, we were able to select sensor combinations for different categorized processes. The collected data were used to compare the performance differences between experienced and new workers through analytical and graphical analyses. Data analyses led us to a series of sample recommendations for novice operators to reduce their learning curves. These recommendations could also improve productivity and minimize production costs and risks related to safety. Given this methodology, manufacturers would be able to generalize the procedure to the majority of manufacturing processes and new sensing technologies in order to capture experts' tacit knowledge. We also used grit blasting as a sample manufacturing environment and five motion sensors to validate our methodology. The selected sensors were able to collect data within the working environment, and with that data we output visualizations and recommendations to the novice.