*Prognostics and Health Management solutions for*
*Reliable Autonomous Systems*
*Description*
The revolution in robotics and autonomous systems (RAS) is unstoppable. The advance of autonomous system applications, such as autonomous transport [1, 2] and autonomous inspections [3], generate multiple benefits for the industry and society, including the improved driving security in autonomous transport, and improved reliability of critical and remote infrastructure through specialized robots and drones.
However, the reliability assurance of RAS is a complex challenge, as it requires incorporating advanced intelligence that should evolve according to run-time operation [4]. The challenging yet exciting, operation context of RAS, hampers the reliability assurance of RAS, which decelerates the acceptance and everyday use of RAS.
Different technological solutions have emerged to improve the design and reliability of RAS [5]. Most of the technological configurations include a combination of mechanical and electrical components, along with onboard software intelligence to adopt decisions without direct human intervention. In this context, using the ever-increasing prognostics and health management solutions, it is possible to develop a prognostics modelling approach for RAS health monitoring using reliability, machine learning, uncertainty modelling and optimization methods [6].
The project’s objective is to develop novel prognostics methods for RAS, which can accurately inform about the model’s confidence in the decisions in real-time and the way to mitigate the existing problem via optimization methods, and always, provide a worst-case estimate on its predictions by using a proper modelling and update of the different sources of uncertainty. In particular, the work will focus on the integration of uncertainties to make prognostic predictions robust, using concepts such as adversarial learning, combined with statistical learning and artificial intelligence methods.
The models developed in this project will be validated with the data collected from industry partners that work with autonomous robots focusing on (i) autonomous remote inspections for renewable energy and (ii) automotive industry.
The project will be developed in <https://www.mondragon.edu/en/home>Mondragon Unibertsitatea within the Electronics and Computer Science Department, in collaboration between the Data Analytics and Signal Processing and Communications groups. Throughout the thesis, the student will engage continuously with industry and stays at different universities and/or research centers will be pursued.
Interested applicants, send your CV and a short motivation letter to: jiaizpurua@mondragon.edu and ezugasti@mondragon.edu
Application deadline. Review of applications will begin July 1st and continue until the position is filled.
*Requirements*
- M.Sc. degree in telecommunications, electronics, computer science, mathematics, embedded systems or a related area.
- Programming skills: Matlab, Python, R, or C++ (samples from prior projects or a GitHub repository are preferred)
- Statistics/mathematics, data science/AI/ML
- Knowledge/experience with autonomous systems is a plus.
- Knowledge/experience with reliability and/or diagnostics/health management methods is a plus.
- Experience with artificial intelligence / optimization methods is a plus.
*References*
[1] Feng, S., Yan, X., Sun, H. et al. Intelligent driving intelligence test for autonomous vehicles with naturalistic and adversarial environment. *Nature Communications* 12, 748 (2021). <https://doi.org/10.1038/s41467-021-21007-8> https://doi.org/10.1038/s41467-021-21007-8
[2] Ellefsen, A. L., Æsøy, V., Ushakov, S., & Zhang, H. (2019). A comprehensive survey of prognostics and health management based on deep learning for autonomous ships. IEEE Transactions on Reliability, 68(2), 720-740
[3] Floreano, D., & Wood, R. J. (2015). Science, technology and the future of small autonomous drones. *nature*, *521*(7553), 460-466.
[4] Aslansefat, K., Kabir, S., Abdullatif, A., Vasudevan, V., & Papadopoulos, Y. (2021). Toward Improving Confidence in Autonomous Vehicle Software: A Study on Traffic Sign Recognition Systems. *Computer*, *54*(8), 66-76.
[5] Elghazel, W., Bahi, J., Guyeux, C., Hakem, M., Medjaher, K., & Zerhouni, N. (2015). Dependability of wireless sensor networks for industrial prognostics and health management. Computers in Industry, 68, 1-15.
[6] Aizpurua, J. I., Catterson, V. M., Papadopoulos, Y., Chiacchio, F., & Manno, G. (2017). Improved dynamic dependability assessment through integration with prognostics. *IEEE Transactions on Reliability*, *66*(3), 893-913.
-- *Joxe Aizpurua.*