Robots Performance Monitoring in Autonomous Manufacturing Operations Using Machine Learning and Big Data

Ahmed Bendaouia, Ahmed Bendaouia, Salma Messaoudi, El Abdelwahed, Jianzhi Li, Jianzhi Li

2025

Abstract

Additive manufacturing has revolutionized industrial automation by enabling flexible and precise production processes. Ensuring the reliability of robotic systems remains a critical challenge. In this study, data-driven approaches are employed to automatically detect faults in the UR5 robot with six joints using Artificial Intelligence. By analyzing sensor data across different combinations of payload, speed, and temperature, this work applies feature engineering and anomaly detection techniques to enhance fault prediction. New features are generated, including binarized anomaly indicators using the interquartile range method and a difference-based time feature to account for the sequential and irregular nature of robot time data. These engineered features allow the use of neural networks (including LSTM), Random Forest, KNN, and GBM models to classify anomalies in position, velocity, and current. A key objective is to evaluate which anomaly type is the most sensitive by analyzing error metrics such as MAE and RMSE, providing insights into the most critical factors affecting robot performance. The experimental findings highlight the superiority of Gradient Boosting Machine and Random Forest in balancing accuracy and computational efficiency, achieving over 99% test accuracy while maintaining short training times. These two models outperform the others, which show a noticeable gap either in training time or test accuracy, demonstrating their effectiveness in improving fault detection and performance monitoring strategies in autonomous experimentation.

Paper Citation

in Harvard Style

Bendaouia A., Messaoudi S., Abdelwahed E. and Li J. (2025). Robots Performance Monitoring in Autonomous Manufacturing Operations Using Machine Learning and Big Data. In Proceedings of the 14th International Conference on Data Science, Technology and Applications - Volume 1: DATA; ISBN 978-989-758-758-0, SciTePress, pages 82-96. DOI: 10.5220/0013526800003967

in Bibtex Style

@conference{data25,
author={Ahmed Bendaouia and Salma Messaoudi and El Abdelwahed and Jianzhi Li},
title={Robots Performance Monitoring in Autonomous Manufacturing Operations Using Machine Learning and Big Data},
booktitle={Proceedings of the 14th International Conference on Data Science, Technology and Applications - Volume 1: DATA},
year={2025},
pages={82-96},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013526800003967},
isbn={978-989-758-758-0},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 14th International Conference on Data Science, Technology and Applications - Volume 1: DATA
TI - Robots Performance Monitoring in Autonomous Manufacturing Operations Using Machine Learning and Big Data
SN - 978-989-758-758-0
AU - Bendaouia A.
AU - Messaoudi S.
AU - Abdelwahed E.
AU - Li J.
PY - 2025
SP - 82
EP - 96
DO - 10.5220/0013526800003967
PB - SciTePress