Increasing Product Safety Through the use of Deep Learning in Manual Assembly
Avoiding product defects takes a high priority in many sectors of industry. An opportunity to reduce errors can be found in the integration of Deep Learning to the production process. Errors might be detected automatically and appropriate interventions could be initiated. The research project investigated to what extent assembly errors can be reduced by using Deep Learning. For this purpose, an Assembly was mounted on two workstations, which only differed in terms of the assistance system used (paper-based instruction vs. digital system with Deep Learning). During this process, all occurring assembly errors were recorded. The results show an error reduction of 45% and prove the high error prevention potential by using Deep Learning.
Bochkovskiy, Alexey; Wang, Chien-Yao; Liao, Hong-Yuan Mark (2020): YOLOv4: Optimal Speed and Accuracy of Object Detection. https://arxiv.org/pdf/2004.10934.pdf [02.02.22].
Redmon, Joseph (2016): Darknet: Open Source Neural Networks in C. http://pjreddie.com/darknet/ [02.02.2022].
Kirwan, Barry (1994): A Guide to Practical Human Reliability Assessment. London: Taylor & Francis.
VDI 4006 Blatt 2 (2015): Menschliche Zuverlässigkeit - Methoden zur quantitativen Bewertung menschlicher Zuverlässigkeit. Düsseldorf: VDI-Verlag.
How to Cite
Conference Proceedings Volume
Copyright (c) 2022 Johanna Gerlach, Alexander Riedel, Frank Engelmann
This work is licensed under a Creative Commons Attribution 4.0 International License.