A data-driven fatigue prediction using recurrent neural networks

Lambay, Arsalan, Liu, Ying ORCID: https://orcid.org/0000-0001-9319-5940, Morgan, Phillip ORCID: https://orcid.org/0000-0002-5672-0758 and Ji, Ze ORCID: https://orcid.org/0000-0002-8968-9902 2021. A data-driven fatigue prediction using recurrent neural networks. Presented at: 3rd International Congress on Human-Computer Interaction, Optimization and Robotic Applications (HORA2021), Ankara, Turkey, 11-13 June 2021. 2021 3rd International Congress on Human-Computer Interaction, Optimization and Robotic Applications (HORA). IEEE, pp. 1-6. 10.1109/HORA52670.2021.9461377

PDF - Accepted Post-Print Version Download (405kB)

Abstract

Industrial revolution 4.0 has marked the era of advances in interaction among machines and humans and cultivate automation. However, manufacturing industries still have tasks which are labor intensive for humans with lots of repetitive actions. These actions along with other factors can cause the worker to be fatigued or exhausted. These in the long term can develop into work-related musculoskeletal disorders (WMSD). Nevertheless, comprehending fatigue in a quantifiable and objective manner is yet an open problem due to the heterogeneity of subjects involved for data collection.In this study a benchmarking dataset comprising of physical fatigue attributes. They are used to perform fatigue prediction for manual material handling task. It includes data collected from Inertial Measurement unit (IMU) and Heart Rate (HR) sensor which is then pre-processed to extract to be used to run the model. The data serves as an input to a time-series prediction model called as Recurrent Neural Network (RNN).

Item Type:	Conference or Workshop Item (Paper)
Date Type:	Published Online
Status:	Published
Schools:	Engineering Psychology
Publisher:	IEEE
ISBN:	9781665440585
Date of First Compliant Deposit:	26 May 2021
Date of Acceptance:	25 May 2021
Last Modified:	28 Nov 2022 12:52
URI:	https://orca.cardiff.ac.uk/id/eprint/141563

Citation Data

Cited 3 times in Scopus. View in Scopus. Powered By Scopus® Data

Actions (repository staff only)

Edit Item