Abstract
In this paper, we study the robustness property of policy optimization (particularly Gauss–Newton gradient descent algorithm which is equivalent to the policy iteration in reinforcement learning) subject to noise at each iteration. By invoking the concept of input-to-state stability and utilizing Lyapunov’s direct method, it is shown that, if the noise is sufficiently small, the policy iteration algorithm converges to a small neighborhood of the optimal solution even in the presence of noise at each iteration. Explicit expressions of the upperbound on the noise and the size of the neighborhood to which the policies ultimately converge are provided. Based on Willems’ fundamental lemma, a learning-based policy iteration algorithm is proposed. The persistent excitation condition can be readily guaranteed by checking the rank of the Hankel matrix related to an exploration signal. The robustness of the learning-based policy iteration to measurement noise and unknown system disturbances is theoretically demonstrated by the input-to-state stability of the policy iteration. Several numerical simulations are conducted to demonstrate the efficacy of the proposed method.
Similar content being viewed by others
Data availability
The data that support the findings of this study are available from the corresponding author, L. Cui, upon reasonable request.
References
Sutton, R. S., & Barto, A. G. (2018). Reinforcement learning: An introduction (2nd ed.). MIT Press.
Fazel, M., Ge, R., Kakade, S., & Mesbahi, M. (2018). Global convergence of policy gradient methods for the linear quadratic regulator. In Proceedings of the 35th international conference on machine learning. vol. 80, pp. 1467–1476.
Bu, J., Mesbahi, A., Fazel, M., & Mesbahi, M. (2019). LQR through the lens of first order methods: Discrete-time case. arXiv:1907.08921 (arXiv e-preprint).
Hu, B., Zhang, K., Li, N., Mesbahi, M., Fazel, M., & Başar, T. (2022). Towards a theoretical foundation of policy optimization for learning control policies. Annual Review of Control, Robotics, and Autonomous Systems, 6(1), 123–158. https://doi.org/10.1146/annurev-control-042920-020021
Mohammadi, H., Zare, A., Soltanolkotabi, M., & Jovanovic, M. R. (2022). Convergence and sample complexity of gradient methods for the model-free linear-quadratic regulator problem. IEEE Transactions on Automatic Control, 67(5), 2435–2450.
Kleinman, D. (1968). On an iterative technique for Riccati equation computations. IEEE Transactions on Automatic Control, 13(1), 114–115. https://doi.org/10.1109/TAC.1968.1098829
Hewer, G. (1971). An iterative technique for the computation of the steady state gains for the discrete optimal regulator. IEEE Transactions on Automatic Control, 16(4), 382–384. https://doi.org/10.1109/TAC.1971.1099755
Bertsekas, D. P. (1995). Dynamic programming and optimal control (Vol. 2). Athena Scientific.
Bertsekas, D. P., & Tsitsiklis, J. N. (1996). Neuro-dynamic programming. Athena Scientific.
Jiang, Y., & Jiang, Z. P. (2012). Computational adaptive optimal control for continuous-time linear systems with completely unknown dynamics. Automatica, 48(10), 2699–2704. https://doi.org/10.1016/j.automatica.2012.06.096
Jiang, Y., & Jiang, Z. P. (2015). Global adaptive dynamic programming for continuous-time nonlinear systems. IEEE Transactions on Automatic Control, 60(11), 2917–2929. https://doi.org/10.1109/TAC.2015.2414811
Cui, L., Pang, B., & Jiang, Z. P. (2023). Learning-based adaptive optimal control of linear time-delay systems: A policy iteration approach. IEEE Transactions on Automatic Control. https://doi.org/10.1109/TAC.2023.3273786
Pang, B., Jiang, Z. P., & Mareels, I. (2020). Reinforcement learning for adaptive optimal control of continuous-time linear periodic systems. Automatica, 118, 109035. https://doi.org/10.1016/j.automatica.2020.109035
Gao, W., & Jiang, Z. P. (2016). Adaptive dynamic programming and adaptive optimal output regulation of linear systems. IEEE Transactions on Automatic Control, 61(12), 4164–4169. https://doi.org/10.1109/TAC.2016.2548662
Pang, B., Cui, L., & Jiang, Z. P. (2022). Human motor learning is robust to control-dependent noise. Biological Cybernetics, 116(12), 307–325.
Liu, T., Cui, L., Pang, B., & Jiang, Z. P. (2021). Data-driven adaptive optimal control of mixed-traffic connected vehicles in a ring road. In 2021 60th IEEE conference on decision and control (CDC), pp. 77–82. https://doi.org/10.1109/CDC45484.2021.9683024.
Cui, L., Ozbay, K., & Jiang, Z. P. (2021) Combined longitudinal and lateral control of autonomous vehicles based on reinforcement learning. In: 2021 American control conference (ACC), pp. 1929–1934. https://doi.org/10.23919/ACC50511.2021.9483388.
Ljung, L. (1998). System identification (pp. 163–173). Birkhäuser. https://doi.org/10.1007/978-1-4612-1768-8_11
Jiang, Z. P., Bian, T., & Gao, W. (2020). Learning-based control: A tutorial and some recent results. Foundations and Trends in Systems and Control, 8(3), 176–284. https://doi.org/10.1561/2600000023
Cui, L., Başar, T., & Jiang, Z. P. (2022). A reinforcement learning look risk-sensitive linear quadratic Gaussian control. In 5th Annual Learning for Dynamics and Control Conference, pp. 534–546.
Pang, B., & Jiang, Z. P. (2023). Reinforcement learning for adaptive optimal stationary control of linear stochastic systems. IEEE Transactions on Automatic Control, 68(4), 2383–2390. https://doi.org/10.1109/TAC.2022.3172250
Cui, L., Wang, S., Zhang, J., Zhang, D., Lai, J., Zheng, Y., Zhang, Z., & Jiang, Z. P. (2021). Learning-based balance control of wheel-legged robots. IEEE Robotics and Automation Letters, 6(4), 7667–7674. https://doi.org/10.1109/LRA.2021.3100269
Sontag, E. (2008). Input to state stability: Basic concepts and results. Lecture notes in mathematics (pp. 163–220). Springer.
Pang, B., & Jiang, Z. P. (2021). Robust reinforcement learning: A case study in linear quadratic regulation. Proceedings of the AAAI Conference on Artificial Intelligence, 35(10), 9303–9311.
Pang, B., Bian, T., & Jiang, Z. P. (2022). Robust policy iteration for continuous-time linear quadratic regulation. IEEE Transactions on Automatic Control, 67(1), 504–511. https://doi.org/10.1109/TAC.2021.3085510
Chen, B. M. (2013). In J. Baillieul & T. Samad (Eds.), H2 optimal control. Springer. https://doi.org/10.1007/978-1-4471-5102-9_204-1
Chen, C.-T. (1999). Linear system theory and design. Oxford University Press.
Mori, T. (1988). Comments on “A matrix inequality associated with bounds on solutions of algebraic Riccati and Lyapunov equation’’ by J. M. Saniuk and I.B. Rhodes. IEEE Transactions on Automatic Control, 33(11), 1088. https://doi.org/10.1109/9.14428
Hespanha, J. P. (2018). Linear systems theory. Princeton University Press.
Zhou, K., Doyle, J. C., & Glover, K. (1996). Robust and optimal control. Prentice Hall.
Willems, J. C., Rapisarda, P., Markovsky, I., & De Moor, B. L. M. (2005). A note on persistency of excitation. Systems and Control Letters, 54(4), 325–329. https://doi.org/10.1016/j.sysconle.2004.09.003
Gahinet, P. M., Laub, A. J., Kenney, C. S., & Hewer, G. A. (1990). Sensitivity of the stable discrete-time Lyapunov equation. IEEE Transactions on Automatic Control, 35(11), 1209–1217. https://doi.org/10.1109/9.59806
Anderson, C. W. (1989). Learning to control an inverted pendulum using neural networks. IEEE Control Systems Magazine, 9(3), 31–37. https://doi.org/10.1109/37.24809
Author information
Authors and Affiliations
Corresponding author
Additional information
This work was supported in part by the National Science Foundation (Nos. ECCS-2210320, CNS-2148304).
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Cui, L., Jiang, ZP. A Lyapunov characterization of robust policy optimization. Control Theory Technol. 21, 374–389 (2023). https://doi.org/10.1007/s11768-023-00163-w
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11768-023-00163-w