Fault-Tolerant Control of ROV Thrusters by using SMC+TD3
DOI:
https://doi.org/10.17979/ja-cea.2026.47.13690Keywords:
Autonomous underwater vehicles, Control architectures in marine systems, Fault-tolerant control, Reinforcement learning control, Sliding mode controlAbstract
This work examines a fault-tolerant control architecture for thruster faults in a BlueROV2-type ROV. The proposed architecture combines a sliding mode control strategy (SMC) with a deep reinforcement learning controller, specifically based on Twin Delayed Deep Deterministic Policy Gradient (TD3). In this design, the SMC provides a robust dynamic control action that remains active at all times, continuously compensating for environmental disturbances, modelling inaccuracies, and other sources of uncertainty. The TD3 agent, once properly trained, provides an additional targeted correction aimed at mitigating the effects of unexpected losses of thruster efficiency. The simulation-based validation shows that the proposed approach reduces yaw error under horizontal thruster faults, while keeping roll and pitch bounded and preserving the nominal SMC behaviour under vertical thruster faults.
References
Benzon, M. V., Sørensen, F. F., Uth, E. T., Jouffroy, J., Liniger, J. y Pedersen, S. (2022). An open-source benchmark simulator: Control of a BlueROV2 underwater robot. Journal of Marine Science and Engineering, vol. 10, no. 12, 1898.
Edwards, C. y Spurgeon, S. K. (1998). Sliding mode control: Theory and applications. Taylor & Francis.
Fossen, T. I. (2021). Handbook of marine craft hydrodynamics and motion control (2nd ed.). John Wiley & Sons.
Goodfellow, I., Bengio, Y. y Courville, A. (2016). Deep learning. MIT Press.
Hosseinnajad, A. y Loueipour, M. (2023). Fault tolerant control system for an ROV based on a novel integral sliding mode control and a state and fault observer in the presence of thruster limitations. Ocean Engineering, vol. 280, 114687. https://doi.org/10.1016/j.oceaneng.2023.114687
Lillicrap, T. P., Hunt, J. J., Pritzel, A., Heess, N., Erez, T., Tassa, Y., Silver, D. y Wierstra, D. (2016). Continuous control with deep reinforcement learning. International Conference on Learning Representations.
Liu, M., Yu, S. y Xu, H. (2025). Fault-Tolerant Control of ROV Based on HJI Integral Sliding Mode. En L. Liu, Y. Niu, W. Fu, y Y. Qu (Eds.), Proceedings of 4th 2024 International Conference on Autonomous Unmanned Systems (4th ICAUS 2024), Springer Nature Singapore, pp. 232–241.
Sainz, J. J., Revestido, E., Llata, J. R., & Velasco, F. J. (2023). Auto-tuning of Non-Linear Model Predictive Controllers for a Remote Operated Vehicle. OCEANS 2023 -- Limerick. https://doi.org/10.1109/OCEANSLimerick52467.2023.10244375
Sainz, J. J., Becerra, V., Revestido, E., Llata, J. R., Alonso, L., & Torre, C. (2024). Adaptive ROV controller for monitoring marine structures. Jornadas de Automatica, 45. https://doi.org/10.17979/ja-cea.2024.45.10807
Sutton, R. S. y Barto, A. G. (2018). Reinforcement learning: An introduction (2nd ed.). MIT Press.
Yin, Z., Gan, W., Ou, J. y Wang, Z. (2024). Fault-Tolerant Control of Remotely Operated Vehicle Thrusters Based on the Dung Beetle Optimizer. 2024 7th International Conference on Advanced Algorithms and Control Engineering (ICAACE), pp. 1186–1189. https://doi.org/10.1109/ICAACE61206.2024.10548829
Downloads
Published
Issue
Section
License
Copyright (c) 2026 José Ramón Llata, Jose Joaquin Sainz, Elías Revestido, Carlos Torre-Ferrero, Luciano Alonso, Sandra Robla
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
