Authors

Z. A. Barhoum, S. A. Kolyubin

Abstract

Modeling the dynamics of underwater robots is a complex task due to the presence of both parametric and functional uncertainties. These arise from interactions with a viscous medium, a priori uncertainty, and variability in the system’s dynamic parameters, as well as the complexity and computational cost of first-principles models and the challenges of identification procedures. This paper proposes the use of neural network parameterization of ordinary differential equations based on the port-Hamiltonian formalism to develop accurate and computationally efficient dynamic models of underwater robots. These models can be used for trajectory prediction, integration with onboard sensor data for localization systems, and controller synthesis. The proposed approach captures both the physical structure of the system and the impact of uncertainties, enabling the creation of physically grounded, data-driven representations of complex nonlinear dynamics. Comparative experiments with classical identification and modeling methods using real-world data from an underwater robot demonstrate advantages of the proposed method in prediction accuracy and its robustness over long-horizon prediction.

Citation

BibTeX

@article{Barhoum_2025,
  title={{Physically consistent dynamic modeling of underwater robots for robust long-horizon motion prediction}},
  volume={68},
  ISSN={0021-3454},
  DOI={10.17586/0021-3454-2025-68-11-983-995},
  number={11},
  journal={Journal of Instrument Engineering},
  publisher={ITMO University},
  author={Barhoum, Z. A. and Kolyubin, S. A.},
  year={2025},
  pages={983--995}
}

Download the bib file

References

  • Choi H-T, Yuh J (2016) Underwater Robots. Springer Handbooks 595–62 – 10.1007/978-3-319-32552-1_25
  • Zhu D., Yan T., Yang S. X. Motion planning and tracking control of unmanned underwater vehicles: technologies, challenges and prospects // arXiv preprint arXiv:2207.04360.2022.
  • Antonelli G, Fossen TI, Yoerger DR (2016) Modeling and Control of Underwater Robots. Springer Handbooks 1285–130 – 10.1007/978-3-319-32552-1_51
  • Zhao Y, Hu Z, Du W, Geng L, Yang Y (2024) Research on Modeling Method of Autonomous Underwater Vehicle Based on a Physics-Informed Neural Network. JMSE 12(5):801. https://doi.org/10.3390/jmse1205080 – 10.3390/jmse12050801
  • Ramirez WA, Kocijan J, Leong ZQ, Nguyen HD, Jayasinghe SG (2021) Dynamic System Identification of Underwater Vehicles Using Multi-Output Gaussian Processes. Int J Autom Comput 18(5):681–693. https://doi.org/10.1007/s11633-021-1308- – 10.1007/s11633-021-1308-x
  • Wu H-M, Karkoub M (2021) Finite-time robust tracking control of an autonomous underwater vehicle in the presence of uncertainties and external current disturbances. Advances in Mechanical Engineering 13(10). https://doi.org/10.1177/1687814021105342 – 10.1177/16878140211053429
  • Kong F, Guo Y, Lyu W (2020) Dynamics Modeling and Motion Control of an New Unmanned Underwater Vehicle. IEEE Access 8:30119–30126. https://doi.org/10.1109/access.2020.297233 – 10.1109/access.2020.2972336
  • Valeriano-Medina Y, Martínez A, Hernández L, Sahli H, Rodríguez Y, Cañizares JR (2013) Dynamic model for an autonomous underwater vehicle based on experimental data. Mathematical and Computer Modelling of Dynamical Systems 19(2):175–200. https://doi.org/10.1080/13873954.2012.71722 – 10.1080/13873954.2012.717226
  • Measurement of an AUV in a Towing Tank. JAFM 12(3):947–959. https://doi.org/10.29252/jafm.12.03.2952 – 10.29252/jafm.12.03.29525
  • {“status”:”error” – 10.1016/j.oceaneng.2008.08.008
  • Faros I., Tanner H. G. System Identification and Adaptive Input Estimation on the Jaiabot Micro Autonomous Underwater Vehicle // arXiv preprint arXiv:2504.02005.2025.
  • Cortez WS, Vasisht S, Tuor A, Koch J, Drgoňa J, Vrabie D (2023) Domain-aware Control-oriented Neural Models for Autonomous Underwater Vehicles. IFAC-PapersOnLine 56(1):228–233. https://doi.org/10.1016/j.ifacol.2023.02.03 – 10.1016/j.ifacol.2023.02.039
  • Sousa CD, Cortesao R (2019) Inertia Tensor Properties in Robot Dynamics Identification: A Linear Matrix Inequality Approach. IEEE/ASME Trans Mechatron 24(1):406–411. https://doi.org/10.1109/tmech.2019.289117 – 10.1109/tmech.2019.2891177
  • Sutanto G. et al. Encoding physical constraints in differentiable newton-euler algorithm // Learning for Dynamics and Control. PMLR, 2020. Р. 804–813.
  • Reuss M. et al. End-to-end learning of hybrid inverse dynamics models for precise and compliant impedance control // arXiv preprint arXiv:2205.13804.2022.
  • Wang J, Le Vang T, Pyrkin AA, Kolyubin SA, Bobtsov AA (2018) Identification of Piecewise Linear Parameters of Regression Models of Non-Stationary Deterministic Systems. Autom Remote Control 79(12):2159–2168. https://doi.org/10.1134/s000511791812006 – 10.1134/s0005117918120068
  • Muñoz F, Cervantes-Rojas JS, Valdovinos JM, Sandre-Hernández O, Salazar S, Romero H (2021) Dynamic Neural Network-Based Adaptive Tracking Control for an Autonomous Underwater Vehicle Subject to Modeling and Parametric Uncertainties. Applied Sciences 11(6):2797. https://doi.org/10.3390/app1106279 – 10.3390/app11062797
  • Szymkowiak M (2025) Using recurrent neural networks to build a data-driven model of an autonomous underwater vehicle. Transportation Research Procedia 83:417–424. https://doi.org/10.1016/j.trpro.2025.03.00 – 10.1016/j.trpro.2025.03.008
  • Singh M., Alexis K. DeepVL: Dynamics and Inertial Measurements-based Deep Velocity Learning for Underwater Odometry // arXiv preprint arXiv:2502.07726.2025.
  • Sun H, Guo W, Lan Y, Wei Z, Gao S, Sun Y, Fu Y (2022) Black-Box Modelling and Prediction of Deep-Sea Landing Vehicles Based on Optimised Support Vector Regression. JMSE 10(5):575. https://doi.org/10.3390/jmse1005057 – 10.3390/jmse10050575
  • Raissi M., Perdikaris P., Karniadakis G. E. Physics informed deep learning (part i): Data-driven solutions of nonlinear partial differential equations // arXiv preprint arXiv:1711.10561.2017.
  • Raissi M., Perdikaris P., Karniadakis G. E. Physics Informed Deep Learning (Part II): Data-driven Discovery of Nonlinear Partial Differential Equations // arXiv e-prints. 2017. Р. arXiv: 1711.10566.
  • Lei L, Gang Y, Jing G (2022) Physics-guided neural network for underwater glider flight modeling. Applied Ocean Research 121:103082. https://doi.org/10.1016/j.apor.2022.10308 – 10.1016/j.apor.2022.103082
  • Greydanus S., Dzamba M., Yosinski J. Hamiltonian neural networks // Advances in neural information processing systems. 2019. Vol. 32.
  • Cranmer M. et al. Lagrangian neural networks // arXiv preprint arXiv:2003.04630.2020.
  • Lutter M., Ritter C., Peters J. Deep lagrangian networks: Using physics as model prior for deep learning // arXiv preprint arXiv:1907.04490.2019.
  • Mirzai B. Physics-informed deep learning for system identification of autonomous underwater vehicles: A Lagrangian neural network approach. 2021.
  • Duong T, Atanasov N (2021) Hamiltonian-based Neural ODE Networks on the SE(3) Manifold For Dynamics Learning and Control. Robotics: Science and Systems XVI – 10.15607/rss.2021.xvii.086
  • Duong T, Altawaitan A, Stanley J, Atanasov N (2024) Port-Hamiltonian Neural ODE Networks on Lie Groups for Robot Dynamics Learning and Control. IEEE Trans Robot 40:3695–3715. https://doi.org/10.1109/tro.2024.342843 – 10.1109/tro.2024.3428433
  • Neary C., Topcu U. Compositional learning of dynamical system models using port-Hamiltonian neural networks // Learning for Dynamics and Control Conference. PMLR, 2023. Р. 679–691.
  • Chen R. T. Q. et al. Neural ordinary differential equations // Advances in neural information processing systems. 2018. Vol. 31.
  • Singh M, Dharmadhikari M, Alexis K (2024) An Online Self-calibrating Refractive Camera Model with Application to Underwater Odometry. 2024 IEEE International Conference on Robotics and Automation (ICRA) 10005–1001 – 10.1109/icra57147.2024.10610643