Prediction of High-Level Actions from the Sequence of Atomic Actions in Assembly Line Workstations | IEEE Conference Publication | IEEE Xplore

Prediction of High-Level Actions from the Sequence of Atomic Actions in Assembly Line Workstations


Abstract:

This study presents a methodology for predicting high-level actions from the sequence of atomic actions in the assembly line workstation. With the rapid development of th...Show More

Abstract:

This study presents a methodology for predicting high-level actions from the sequence of atomic actions in the assembly line workstation. With the rapid development of the automotive industry, changes in product design are becoming more frequent. These situations require quick line balancing in response to product changes to maximize assembly line productivity. One of the key components of line balancing is cycle time analysis, which tracks the time required to complete individual operations within and across the stations. Sensors such as IMU, microphones, and pressure can be used to recognize atomic operations; however, it is difficult to recognize user-level or high-level activities directly from them due to the complexity involved in recognizing these advanced actions. To fill this gap, especially in assembly line workstations where every workstation has a pre-defined sequence of high-level actions that workers perform repeatedly, we propose a methodology that takes the sequence of atomic actions and the workstation's structure manual as input and predicts the high-level actions through sequence matching. The developed approach has achieved an average of 50.92 % accuracy, whereas the baseline method could only achieve an average of 23.48 % accuracy when evaluated using the Assembly 101 dataset. These results demonstrate the superiority of the proposed approach for predicting high-level action from the sequence of atomic actions.
Date of Conference: 10-13 September 2024
Date Added to IEEE Xplore: 16 October 2024
ISBN Information:

ISSN Information:

Conference Location: Padova, Italy

References

References is not available for this document.