Fault Prediction and Estimation of Automotive LiDAR Signals Using Transfer Learning-Based Domain Generalization

##plugins.themes.bootstrap3.article.main##

##plugins.themes.bootstrap3.article.sidebar##

Published Jun 27, 2024
Sanghoon Lee Jaewook Lee Jongsoo Lee

Abstract

Autonomous vehicles (AVs) are undergoing level 4 technology development and should have a system that can be operated without driver’s intervention, so that it must be possible to diagnose failures and predict life cycle themselves. In this study, we propose a technology to estimate signal changes and sensor faults through transfer learning-based domain generalization (TLDG) using limited actual vehicle test information from LiDAR for autonomous vehicles. Because autonomous vehicles operate in various climate/weather conditions over the world, their mechanical, electrical and electronic components must also have stable performance in all environmental conditions. However, an electronic device, especially laser diode (LD), which is one of core components of LiDAR, shows various degradation aspects depending on environmental conditions. We acquired multivariate LiDAR performance data under various environmental conditions through an actual vehicle test driving of about 2,000 km in summer and winter, and based on this, we create the LiDAR fault diagnosis and performance prediction model generalized to the domain under various environmental conditions. Fault prediction and estimation model created through summer and winter data in the environment domain will also adapt to other environmental conditions such as spring and fall. To develop highly accurate performance estimation models under various environmental conditions based on limited data, it is very important to extract correlations and characteristics between data, including environmental conditions. We employ the data augmentation techniques to solve the problem of lack of training data and apply bi-directional Bayesian transfer learning to generalize data and models under uncertainty. To prove the effectiveness of the present study, the data from actual vehicle tests conducted at different temperatures will be divided into train data and test data, and the validity of the generalized degradation performance estimation model will be statistically validated. The proposed domain generalization method, i.e., TLDG can be utilized to estimate signal changes and sensor faults in LiDAR under unexperienced environmental conditions such as weather changes, and even freezing and hot regions.

How to Cite

Lee, S., Lee, J., & Lee, J. (2024). Fault Prediction and Estimation of Automotive LiDAR Signals Using Transfer Learning-Based Domain Generalization. PHM Society European Conference, 8(1), 6. https://doi.org/10.36001/phme.2024.v8i1.4027
Abstract 177 | PDF Downloads 117

##plugins.themes.bootstrap3.article.details##

Keywords

Autonomous vehicle, LiDAR, Fault diagnosis, Domain generalization

References
Zhao, X., Fang, Y., Min, H., Wu, X., Wang, W., & Teixeira, R. (2023). Potential sources of sensor data anomalies for autonomous vehicles: An overview from road vehicle safety perspective. Expert Systems with Applications, 121358.
Gültekin, Ö., Cinar, E., Özkan, K., & Yazıcı, A. (2022). Real-time fault detection and condition monitoring for industrial autonomous transfer vehicles utilizing edge artificial intelligence. Sensors, 22(9), 3208.
Ahmed, M., Mahmood, A. N., & Hu, J. (2016). A survey of network anomaly detection techniques. Journal of Network and Computer Applications, 60, 19-31.
Fang, Y., Min, H., Wu, X., Lei, X., Chen, S., Teixeira, R., & Zhao, X., (2023). Toward interpretability in fault diagnosis for autonomous vehicles: interpretation of sensor data anomalies. IEEE Sensors Journal, 23(5), 5014-5027.
Xu, Y., Sun, Y., Liu, X., & Zheng, Y. (2019). A digital-twin-assisted fault diagnosis using deep transfer learning. IEEE Access, 7, 19990-19999.
Li, X., Zhang, W., Ma, H., Luo, Z., & Li, X. (2020). Domain generalization in rotating machinery fault diagnostics using deep neural networks. Neurocomputing, 403, 409-420.
Li, Q., Chen, L., Kong, L., Wang, D., Xia, M., & Shen, C. (2023). Cross-domain augmentation diagnosis: An adversarial domain-augmented generalization method for fault diagnosis under unseen working conditions. Reliability Engineering & System Safety, 234, 109171.
Chang, F., Jafarzadeh, E., Del Gatto, J., Cran, G., & Sadjadi, H. (2023, October). Failure Mode Investigation to Enable LiDAR Health Monitoring for Automotive Application. Annual Conference of the PHM Society, Vol. 15, No. 1
Yuan, H., Cui, N., Li, C., Cui, Z., & Chang, L., (2023). Early stage internal short circuit fault diagnosis for lithium-ion batteries, Journal of Energy Storage, Vol. 57, 106196
Tang, L., Tian, H., Huang, H., Shi, S., & Ji, Q., (2022). A survey of mechanical fault diagnosis based on audio signal analysis, Measurement, Vol. 220, 113294
Weiss, K., Khoshgoftaar, T. M., & Wang, D., (2016). A survey of transfer learning, Journal of Big data, 3, 1-40.
Section
Technical Papers