Plug-and-Play Physics-Informed Learning Using Uncertainty Quantified Port-Hamiltonian Models

Authors

Kaiyuan Tan, Peilun Li, Jun Wang, Thomas Beckers

Abstract

The ability to predict trajectories of surrounding agents and obstacles is a crucial component in many robotic applications. Data-driven approaches are commonly adopted for state prediction in scenarios where the underlying dynamics are unknown. However, the performance, reliability, and uncertainty of data-driven predictors become compromised when encountering out-of-distribution observations relative to the training data. In this paper, we introduce a Plug-and-Play Physics-Informed Machine Learning (PnP-PIML) framework to address this challenge. Our method employs conformal prediction to identify outlier dynamics and, in that case, switches from a nominal predictor to a physics-consistent model, namely distributed Port-Hamiltonian systems (dPHS). We leverage Gaussian processes to model the energy function of the dPHS, enabling not only the learning of system dynamics but also the quantification of predictive uncertainty through its Bayesian nature. In this way, the proposed framework produces reliable physics-informed predictions even for the out-of-distribution scenarios.

Citation

• Journal: 2025 IEEE International Conference on Robotics and Automation (ICRA)
• Year: 2025
• Volume:
• Issue:
• Pages: 10980–10986
• Publisher: IEEE
• DOI: 10.1109/icra55743.2025.11127683

BibTeX

@inproceedings{Tan_2025,
  title={{Plug-and-Play Physics-Informed Learning Using Uncertainty Quantified Port-Hamiltonian Models}},
  DOI={10.1109/icra55743.2025.11127683},
  booktitle={{2025 IEEE International Conference on Robotics and Automation (ICRA)}},
  publisher={IEEE},
  author={Tan, Kaiyuan and Li, Peilun and Wang, Jun and Beckers, Thomas},
  year={2025},
  pages={10980--10986}
}

Download the bib file

References

• Karabegović, I., Karabegović, E., Mahmić, M. & Husak, E. The application of service robots for logistics in manufacturing processes. Adv produc engineer manag 10, 185–194 (2015) – 10.14743/apem2015.4.201
• Vukobratovic, M. Robot-environment dynamic interaction survey and future trends. J. Comput. Syst. Sci. Int. 49, 329–342 (2010) – 10.1134/s1064230710020188
• Hassanein, A., Elhawary, M., Jaber, N. & El-Abd, M. An autonomous firefighting robot. 2015 International Conference on Advanced Robotics (ICAR) 530–535 (2015) doi:10.1109/icar.2015.7251507 – 10.1109/icar.2015.7251507
• Lydia, M., Suresh Kumar, S., Immanuel Selvakumar, A. & Edwin Prem Kumar, G. Linear and non-linear autoregressive models for short-term wind speed forecasting. Energy Conversion and Management 112, 115–124 (2016) – 10.1016/j.enconman.2016.01.007
• Wei, Z., Chen, W., Wang, H. & Wang, J. Manipulator motion planning using flexible obstacle avoidance based on model learning. International Journal of Advanced Robotic Systems 14, 172988141770393 (2017) – 10.1177/1729881417703930
• Long, Z., Lu, Y. & Dong, B. PDE-Net 2.0: Learning PDEs from data with a numeric-symbolic hybrid deep network. Journal of Computational Physics 399, 108925 (2019) – 10.1016/j.jcp.2019.108925
• Stephany, R. & Earls, C. PDE-READ: Human-readable partial differential equation discovery using deep learning. Neural Networks 154, 360–382 (2022) – 10.1016/j.neunet.2022.07.008
• Hou, Z.-S. & Wang, Z. From model-based control to data-driven control: Survey, classification and perspective. Information Sciences 235, 3–35 (2013) – 10.1016/j.ins.2012.07.014
• Zhang, Q., Hu, S., Sun, J., Chen, Q. A. & Mao, Z. M. On Adversarial Robustness of Trajectory Prediction for Autonomous Vehicles. 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 15138–15147 (2022) doi:10.1109/cvpr52688.2022.01473 – 10.1109/cvpr52688.2022.01473
• Tan, Targeted adversarial attacks against neural network trajectory predictors. Learning for Dynamics and Control Conference. PMLR (2023)
• Raissi, M., Perdikaris, P. & Karniadakis, G. E. Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations. Journal of Computational Physics 378, 686–707 (2019) – 10.1016/j.jcp.2018.10.045
• Si, C. & Yan, M. Initialization-enhanced physics-informed neural network with domain decomposition (IDPINN). Journal of Computational Physics 530, 113914 (2025) – 10.1016/j.jcp.2025.113914
• Nageshrao, S. P., Lopes, G. A. D., Jeltsema, D. & Babuska, R. Port-Hamiltonian Systems in Adaptive and Learning Control: A Survey. IEEE Trans. Automat. Contr. 61, 1223–1238 (2016) – 10.1109/tac.2015.2458491
• Duong, T. & Atanasov, N. Hamiltonian-based Neural ODE Networks on the SE(3) Manifold For Dynamics Learning and Control. Robotics: Science and Systems XVII (2021) doi:10.15607/rss.2021.xvii.086 – 10.15607/rss.2021.xvii.086
• Beckers, T., Seidman, J., Perdikaris, P. & Pappas, G. J. Gaussian Process Port-Hamiltonian Systems: Bayesian Learning with Physics Prior. 2022 IEEE 61st Conference on Decision and Control (CDC) 1447–1453 (2022) doi:10.1109/cdc51059.2022.9992733 – 10.1109/cdc51059.2022.9992733
• Li, P., Tan, K. & Beckers, T. PyGpPHs: A Python Package for Bayesian Modeling of Port-Hamiltonian Systems. IFAC-PapersOnLine 58, 54–59 (2024) – 10.1016/j.ifacol.2024.08.256
• Tan, Physics-constrained learning for PDE systems with uncertainty quantified portHamiltonian models. 6th Annual Learning for Dynamics & Control Conference
• Angelopoulos, A gentle introduction to conformal prediction and distribution-free uncertainty quantification. arXiv preprint (2021)
• Zhao, Y., Hoxha, B., Fainekos, G., Deshmukh, J. V. & Lindemann, L. Robust Conformal Prediction for STL Runtime Verification under Distribution Shift. 2024 ACM/IEEE 15th International Conference on Cyber-Physical Systems (ICCPS) 169–179 (2024) doi:10.1109/iccps61052.2024.00022 – 10.1109/iccps61052.2024.00022
• Macchelli, A., van der Schaft, A. J. & Melchiorri, C. Port Hamiltonian formulation of infinite dimensional systems I. Modeling. 2004 43rd IEEE Conference on Decision and Control (CDC) (IEEE Cat. No.04CH37601) 3762-3767 Vol.4 (2004) doi:10.1109/cdc.2004.1429324 – 10.1109/cdc.2004.1429324
• Van der Schaft, L2-gain and passivity techniques in nonlinear control. Springer (2000)
• Jidling, Linearly constrained Gaussian processes. Advances in Neural Information Processing Systems (2017)