Port-Hamiltonian Neural ODE Networks on Lie Groups for Robot Dynamics Learning and Control

Authors

Thai Duong, Abdullah Altawaitan, Jason Stanley, Nikolay Atanasov

Abstract

Accurate models of robot dynamics are critical for safe and stable control and generalization to novel operational conditions. Hand-designed models, however, may be insufficiently accurate, even after careful parameter tuning. This motivates the use of machine learning techniques to approximate the robot dynamics over a training set of state-control trajectories. The dynamics of many robots are described in terms of their generalized coordinates on a matrix Lie group, e.g., on \( ext{SE}(3) \) for ground, aerial, and underwater vehicles, and generalized velocity, and satisfy conservation of energy principles. This article proposes a port-Hamiltonian formulation over a Lie group of the structure of a neural ordinary differential equation (ODE) network to approximate the robot dynamics. In contrast to a black-box ODE network, our formulation embeds energy conservation principle and Lie group’s constraints in the dynamics model and explicitly accounts for energy-dissipation effect such as friction and drag forces in the dynamics model. We develop energy shaping and damping injection control for the learned, potentially under-actuated Hamiltonian dynamics to enable a unified approach for stabilization and trajectory tracking with various robot platforms.

Citation

• Journal: IEEE Transactions on Robotics
• Year: 2024
• Volume: 40
• Issue:
• Pages: 3695–3715
• Publisher: Institute of Electrical and Electronics Engineers (IEEE)
• DOI: 10.1109/tro.2024.3428433

BibTeX

@article{Duong_2024,
  title={{Port-Hamiltonian Neural ODE Networks on Lie Groups for Robot Dynamics Learning and Control}},
  volume={40},
  ISSN={1941-0468},
  DOI={10.1109/tro.2024.3428433},
  journal={IEEE Transactions on Robotics},
  publisher={Institute of Electrical and Electronics Engineers (IEEE)},
  author={Duong, Thai and Altawaitan, Abdullah and Stanley, Jason and Atanasov, Nikolay},
  year={2024},
  pages={3695--3715}
}

Download the bib file

References

• Ljung, L. System Identification. Wiley Encyclopedia of Electrical and Electronics Engineering (1999) doi:10.1002/047134608x.w1046 – 10.1002/047134608x.w1046
• Nguyen-Tuong, D. & Peters, J. Model learning for robot control: a survey. Cognitive Processing vol. 12 319–340 (2011) – 10.1007/s10339-011-0404-1
• Deisenroth, PILCO: A model-based and data-efficient approach to policy search. Proc. Int. Conf. Mach. Learn. (2011)
• Williams, G. et al. Information theoretic MPC for model-based reinforcement learning. 2017 IEEE International Conference on Robotics and Automation (ICRA) 1714–1721 (2017) doi:10.1109/icra.2017.7989202 – 10.1109/icra.2017.7989202
• Raissi, Multistep neural networks for data-driven discovery of nonlinear dynamical systems. (2018)
• Chua, Deep reinforcement learning in a handful of trials using probabilistic dynamics models. Proc. Adv. Neural Inf. Process. Syst. (2018)
• Lutter, M. & Peters, J. Combining physics and deep learning to learn continuous-time dynamics models. The International Journal of Robotics Research vol. 42 83–107 (2023) – 10.1177/02783649231169492
• Gupta, A general framework for structured learning of mechanical systems. (2019)
• Cranmer, Lagrangian neural networks. Proc. ICLR Workshop Integration Deep Neural Models Differ. Equ. (2020)
• Greydanus, Hamiltonian neural networks. Proc. Adv. Neural Inf. Process. Syst. (2019)
• Chen, Symplectic recurrent neural networks. Proc. Int. Conf. Learn. Representations (2020)
• Roehrl, M. A., Runkler, T. A., Brandtstetter, V., Tokic, M. & Obermayer, S. Modeling System Dynamics with Physics-Informed Neural Networks Based on Lagrangian Mechanics. IFAC-PapersOnLine vol. 53 9195–9200 (2020) – 10.1016/j.ifacol.2020.12.2182
• Lu, ModLaNets: Learning generalisable dynamics via modularity and physical inductive bias. Proc. Int. Conf. Mach. Learn. (2022)
• Neary, Compositional learning of dynamical system models using port-Hamiltonian neural networks. Proc. Learn. Dyn. Control Conf. (2023)
• Sanchez-Gonzalez, Graph networks as learnable physics engines for inference and control. Proc. Int. Conf. Mach. Learn. (2018)
• Lutter, Deep lagrangian networks: Using physics as model prior for deep learning. Proc. Int. Conf. Learn. Representations (2018)
• Hall, B. C. Lie Groups, Lie Algebras, and Representations. Graduate Texts in Mathematics 333–366 (2013) doi:10.1007/978-1-4614-7116-5_16 – 10.1007/978-1-4614-7116-5_16
• Lynch, K. M. & Park, F. C. Modern Robotics. (2017) doi:10.1017/9781316661239 – 10.1017/9781316661239
• Falorsi, Neural ordinary differential equations on manifolds. (2020)
• Elamvazhuthi, Learning on manifolds: Universal approximations properties using geometric controllability conditions for neural ODEs. Proc. Learn. Dyn. Control Conf. (2023)
• Lou, Neural manifold ordinary differential equations. Proc. Adv. Neural Inf. Process. Syst. (2020)
• Wotte, Y. P., Califano, F. & Stramigioli, S. Optimal potential shaping on SE(3) via neural ordinary differential equations on Lie groups. The International Journal of Robotics Research vol. 43 2221–2244 (2024) – 10.1177/02783649241256044
• Lurie, A. I. Analytical Mechanics. (Springer Berlin Heidelberg, 2002). doi:10.1007/978-3-540-45677-3 – 10.1007/978-3-540-45677-3
• Holm, D. D. Geometric Mechanics. (IMPERIAL COLLEGE PRESS, 2008). doi:10.1142/p557 – 10.1142/p557
• Bertalan, T., Dietrich, F., Mezić, I. & Kevrekidis, I. G. On learning Hamiltonian systems from data. Chaos: An Interdisciplinary Journal of Nonlinear Science vol. 29 (2019) – 10.1063/1.5128231
• Finzi, Simplifying Hamiltonian and Lagrangian neural networks via explicit constraints. Proc. Adv. Neural Inf. Process. Syst. (2020)
• Zhong, Symplectic ODE-Net: Learning Hamiltonian dynamics with control. Proc. Int. Conf. Learn. Representations (2019)
• Willard, J., Jia, X., Xu, S., Steinbach, M. & Kumar, V. Integrating Scientific Knowledge with Machine Learning for Engineering and Environmental Systems. ACM Computing Surveys vol. 55 1–37 (2022) – 10.1145/3514228
• Maciejewski, A. J. Hamiltonian formalism for Euler parameters. Celestial Mechanics vol. 37 47–57 (1985) – 10.1007/bf01230340
• Shivarama, R. & Fahrenthold, E. P. Hamilton’s Equations With Euler Parameters for Rigid Body Dynamics Modeling. Journal of Dynamic Systems, Measurement, and Control vol. 126 124–130 (2004) – 10.1115/1.1649977
• Chen, Neural ordinary differential equations. Proc. Adv. Neural Inf. Process. Syst. (2018)
• 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
• Lee, Global Formulations of Lagrangian and Hamiltonian Dynamics on Manifolds (2017)
• Zhong, Dissipative SymODEN: Encoding Hamiltonian dynamics with dissipation and control into deep learning. Proc. ICLR Workshop Integration Deep Neural Models Differ. Equ. (2020)
• Song, Y., Romero, A., Müller, M., Koltun, V. & Scaramuzza, D. Reaching the limit in autonomous racing: Optimal control versus reinforcement learning. Science Robotics vol. 8 (2023) – 10.1126/scirobotics.adg1462
• Salzmann, T. et al. Real-Time Neural MPC: Deep Learning Model Predictive Control for Quadrotors and Agile Robotic Platforms. IEEE Robotics and Automation Letters vol. 8 2397–2404 (2023) – 10.1109/lra.2023.3246839
• Scaramuzza, D. & Kaufmann, E. Learning Agile, Vision-Based Drone Flight: From Simulation to Reality. Springer Proceedings in Advanced Robotics 11–18 (2023) doi:10.1007/978-3-031-25555-7_2 – 10.1007/978-3-031-25555-7_2
• Loquercio, A. et al. Learning high-speed flight in the wild. Science Robotics vol. 6 (2021) – 10.1126/scirobotics.abg5810
• Ibarz, J. et al. How to train your robot with deep reinforcement learning: lessons we have learned. The International Journal of Robotics Research vol. 40 698–721 (2021) – 10.1177/0278364920987859
• Nghiem, T. X. et al. Physics-Informed Machine Learning for Modeling and Control of Dynamical Systems. 2023 American Control Conference (ACC) 3735–3750 (2023) doi:10.23919/acc55779.2023.10155901 – 10.23919/acc55779.2023.10155901
• Djeumou, Neural networks with physics-informed architectures and constraints for dynamical systems modeling. Proc. Learn. Dyn. Control Conf. (2022)
• Thorpe, A. J., Neary, C., Djeumou, F., K. Oishi, M. M. & Topcu, U. Physics-Informed Kernel Embeddings: Integrating Prior System Knowledge with Data-Driven Control. 2024 American Control Conference (ACC) 3130–3137 (2024) doi:10.23919/acc60939.2024.10644707 – 10.23919/acc60939.2024.10644707
• Ruthotto, L. & Haber, E. Deep Neural Networks Motivated by Partial Differential Equations. Journal of Mathematical Imaging and Vision vol. 62 352–364 (2019) – 10.1007/s10851-019-00903-1
• Wang, Incorporating symmetry into deep dynamics models for improved generalization. Proc. Int. Conf. Learn. Representations (2021)
• Lutter, M., Listmann, K. & Peters, J. Deep Lagrangian Networks for end-to-end learning of energy-based control for under-actuated systems. 2019 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) 7718–7725 (2019) doi:10.1109/iros40897.2019.8968268 – 10.1109/iros40897.2019.8968268
• 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
• Beckers, T. Data-Driven Bayesian Control of Port-Hamiltonian Systems. 2023 62nd IEEE Conference on Decision and Control (CDC) (2023) doi:10.1109/cdc49753.2023.10384219 – 10.1109/cdc49753.2023.10384219
• Leimkuhler, Simulating Hamiltonian Dynamics (2004)
• Toth, Hamiltonian generative networks. Proc. Int. Conf. Learn. Representations (2019)
• Mason, Learning interpretable dynamics from images of a freely rotating 3D rigid body. Proc. Third Symp. Knowl. guided Mach. Learn. (2022)
• van der Schaft, A. & Jeltsema, D. Port-Hamiltonian Systems Theory: An Introductory Overview. Foundations and Trends® in Systems and Control vol. 1 173–378 (2014) – 10.1561/2600000002
• Marsden, J. E. & West, M. Discrete mechanics and variational integrators. Acta Numerica vol. 10 357–514 (2001) – 10.1017/s096249290100006x
• Havens, Forced variational integrator networks for prediction and control of mechanical systems. Proc. Learn. Dyn. Control (2021)
• Duruisseaux, Lie group forced variational integrator networks for learning and control of robot systems. Proc. Learn. Dyn. Control Conf. (2023)
• Chen, Data-driven prediction of general Hamiltonian dynamics via learning exactly-symplectic maps. Proc. Int. Conf. Mach. Learn. (2021)
• So, Data-driven discovery of non-newtonian astronomy via learning non-euclidean hamiltonian. Proc. ICML Mach. Learn. Phys. Sci. Workshop (2022)
• Borrelli, F., Bemporad, A. & Morari, M. Predictive Control for Linear and Hybrid Systems. (2017) doi:10.1017/9781139061759 – 10.1017/9781139061759
• Ortega, R., Spong, M. W., Gomez-Estern, F. & Blankenstein, G. Stabilization of a class of underactuated mechanical systems via interconnection and damping assignment. IEEE Transactions on Automatic Control vol. 47 1218–1233 (2002) – 10.1109/tac.2002.800770
• Acosta, J. A., Sanchez, M. I. & Ollero, A. Robust control of underactuated Aerial Manipulators via IDA-PBC. 53rd IEEE Conference on Decision and Control 673–678 (2014) doi:10.1109/cdc.2014.7039459 – 10.1109/cdc.2014.7039459
• Cieza, O. B. & Reger, J. IDA-PBC for Underactuated Mechanical Systems in Implicit Port-Hamiltonian Representation. 2019 18th European Control Conference (ECC) (2019) doi:10.23919/ecc.2019.8795994 – 10.23919/ecc.2019.8795994
• Wang, Z. & Goldsmith, P. Modified energy-balancing-based control for the tracking problem. IET Control Theory & Applications vol. 2 310–322 (2008) – 10.1049/iet-cta:20070124
• Souza, C., Raffo, G. V. & Castelan, E. B. Passivity Based Control of a Quadrotor UAV. IFAC Proceedings Volumes vol. 47 3196–3201 (2014) – 10.3182/20140824-6-za-1003.02335
• Furieri, Distributed neural network control with dependability guarantees: A compositional port-Hamiltonian approach. Proc. Learn. Dyn. Control Conf. (2022)
• Galimberti, C. L., Furieri, L., Xu, L. & Ferrari-Trecate, G. Hamiltonian Deep Neural Networks Guaranteeing Nonvanishing Gradients by Design. IEEE Transactions on Automatic Control vol. 68 3155–3162 (2023) – 10.1109/tac.2023.3239430
• Sebastián, E., Duong, T., Atanasov, N., Montijano, E. & Sagüés, C. LEMURS: Learning Distributed Multi-Robot Interactions. 2023 IEEE International Conference on Robotics and Automation (ICRA) 7713–7719 (2023) doi:10.1109/icra48891.2023.10161328 – 10.1109/icra48891.2023.10161328
• Marsden, J. E. & Ratiu, T. S. Introduction to Mechanics and Symmetry. Texts in Applied Mathematics (Springer New York, 1994). doi:10.1007/978-1-4612-2682-6 – 10.1007/978-1-4612-2682-6
• Zhou, Y., Barnes, C., Lu, J., Yang, J. & Li, H. On the Continuity of Rotation Representations in Neural Networks. 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 5738–5746 (2019) doi:10.1109/cvpr.2019.00589 – 10.1109/cvpr.2019.00589
• Barfoot, T. D. State Estimation for Robotics. (2017) doi:10.1017/9781316671528 – 10.1017/9781316671528
• Hemingway, E. G. & O’Reilly, O. M. Perspectives on Euler angle singularities, gimbal lock, and the orthogonality of applied forces and applied moments. Multibody System Dynamics vol. 44 31–56 (2018) – 10.1007/s11044-018-9620-0
• Faessler, M., Franchi, A. & Scaramuzza, D. Differential Flatness of Quadrotor Dynamics Subject to Rotor Drag for Accurate Tracking of High-Speed Trajectories. IEEE Robotics and Automation Letters vol. 3 620–626 (2018) – 10.1109/lra.2017.2776353
• Forni, P., Jeltsema, D. & Lopes, G. A. D. Port-Hamiltonian Formulation of Rigid-Body Attitude Control. IFAC-PapersOnLine vol. 48 164–169 (2015) – 10.1016/j.ifacol.2015.10.233
• Rashad, R., Califano, F. & Stramigioli, S. Port-Hamiltonian Passivity-Based Control on SE(3) of a Fully Actuated UAV for Aerial Physical Interaction Near-Hovering. IEEE Robotics and Automation Letters vol. 4 4378–4385 (2019) – 10.1109/lra.2019.2932864
• Delmerico, J. & Scaramuzza, D. A Benchmark Comparison of Monocular Visual-Inertial Odometry Algorithms for Flying Robots. 2018 IEEE International Conference on Robotics and Automation (ICRA) 2502–2509 (2018) doi:10.1109/icra.2018.8460664 – 10.1109/icra.2018.8460664
• Mohamed, S. A. S. et al. A Survey on Odometry for Autonomous Navigation Systems. IEEE Access vol. 7 97466–97486 (2019) – 10.1109/access.2019.2929133
• Paszke, PyTorch: An imperative style, high-performance deep learning library. Proc. Adv. Neural Inf. Process. Syst. (2019)
• Dormand, J. R. & Prince, P. J. A family of embedded Runge-Kutta formulae. Journal of Computational and Applied Mathematics vol. 6 19–26 (1980) – 10.1016/0771-050x(80)90013-3
• Lienen, torchode: A parallel ODE solver for PyTorch. Proc. Symbiosis Deep Learn. Differ. Equ. II (2022)
• Iserles, A., Munthe-Kaas, H. Z., Nørsett, S. P. & Zanna, A. Lie-group methods. Acta Numerica vol. 9 215–365 (2000) – 10.1017/s0962492900002154
• Lee, T., Leok, M. & McClamroch, N. H. Geometric tracking control of a quadrotor UAV on SE(3). 49th IEEE Conference on Decision and Control (CDC) 5420–5425 (2010) doi:10.1109/cdc.2010.5717652 – 10.1109/cdc.2010.5717652
• Khalil, Nonlinear Systems (2002)
• Panerati, Learning to fly: A PyBullet gym environment to learn the control of multiple nano-quadcopters. (2020)
• Meier, L., Honegger, D. & Pollefeys, M. PX4: A node-based multithreaded open source robotics framework for deeply embedded platforms. 2015 IEEE International Conference on Robotics and Automation (ICRA) 6235–6240 (2015) doi:10.1109/icra.2015.7140074 – 10.1109/icra.2015.7140074
• Boumal, N. An Introduction to Optimization on Smooth Manifolds. (2023) doi:10.1017/9781009166164 – 10.1017/9781009166164
• Arathoon, Coadjoint orbits of the special Euclidean group. (2015)