Learning Min-norm Stabilizing Control Laws for Systems with Unknown Dynamics

被引:0
|
作者
Westenbroek, Tyler [1 ]
Castaneda, Fernando [2 ]
Agrawal, Ayush [2 ]
Sastry, S. Shankar [1 ]
Sreenath, Koushil [2 ]
机构
[1] Univ Calif Berkeley, Dept Elect Engn & Comp Sci, Berkeley, CA 94720 USA
[2] Univ Calif Berkeley, Dept Mech Engn, Berkeley, CA 94720 USA
来源
2020 59TH IEEE CONFERENCE ON DECISION AND CONTROL (CDC) | 2020年
基金
美国国家科学基金会;
关键词
TO-STATE STABILITY; ADAPTIVE-CONTROL; LYAPUNOV;
D O I
10.1109/cdc42340.2020.9304118
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
This paper introduces a framework for learning a minimum-norm stabilizing controller for a system with unknown dynamics using model-free policy optimization methods. The approach begins by first designing a Control Lyapunov Function (CLF) for a (possibly inaccurate) dynamics model for the system, along with a function which specifies a minimum acceptable rate of energy dissipation for the CLF at different points in the state-space. Treating the energy dissipation condition as a constraint on the desired closed-loop behavior of the real-world system, we use penalty methods to formulate an unconstrained optimization problem over the parameters of a learned controller, which can be solved using model-free policy optimization algorithms using data collected from the plant. We discuss when the optimization learns a stabilizing controller for the real world system and derive conditions on the structure of the learned controller which ensure that the optimization is strongly convex, meaning the globally optimal solution can be found reliably. We validate the approach in simulation, first for a double pendulum, and then generalize the framework to learn stable walking controllers for underactuated bipedal robots using the Hybrid Zero Dynamics framework. By encoding a large amount of structure into the learning problem, we are able to learn stabilizing controllers for both systems with only minutes or even seconds of training data.
引用
收藏
页码:737 / 744
页数:8
相关论文
共 50 条
  • [21] Learning control of flexible manipulator with unknown dynamics
    Chen, Zhiguang
    Yang, Chenguang
    Liu, Xin
    Wang, Min
    ASSEMBLY AUTOMATION, 2017, 37 (03) : 304 - 313
  • [22] Reinforcement Q-Learning for PDF Tracking Control of Stochastic Systems with Unknown Dynamics
    Yang, Weiqing
    Zhou, Yuyang
    Zhang, Yong
    Ren, Yan
    MATHEMATICS, 2024, 12 (16)
  • [23] LEARNING THE DYNAMICS FOR UNKNOWN HYPERBOLIC CONSERVATION LAWS USING DEEP NEURAL NETWORKS
    Chen, Zhen
    Gelb, Anne
    Lee, Yoonsang
    SIAM JOURNAL ON SCIENTIFIC COMPUTING, 2024, 46 (02): : A825 - A850
  • [24] Robust Iterative Learning Control Laws with Full Dynamics
    Hladowski, Lukasz
    Paszke, Wojciech
    Galkowski, Krzysztof
    Rogers, Eric
    2016 IEEE 55TH CONFERENCE ON DECISION AND CONTROL (CDC), 2016, : 2370 - 2375
  • [25] MIN-MAX CONTROL AND LINEAR FEEDBACK FOR DYNAMIC-SYSTEMS APPROXIMATED IN NORM
    NEGRO, A
    MILANESE, M
    REVUE FRANCAISE D AUTOMATIQUE INFORMATIQUE RECHERCHE OPERATIONNELLE, 1976, 10 (03): : 61 - 80
  • [26] Stabilizing Control of a Class of Unknown Nonlinear Systems Using Dynamic Neural Networks
    Farid, Farshad
    Pourboghrat, Farzad
    2010 AMERICAN CONTROL CONFERENCE, 2010, : 4919 - 4924
  • [27] Design of stabilizing control laws for mechanical systems based on Lyapunov's method
    Ohtsuka, T
    Fujii, HA
    JOURNAL OF GUIDANCE CONTROL AND DYNAMICS, 1996, 19 (01) : 172 - 180
  • [28] Robust stabilizing control laws for a class of second-order switched systems
    Hu, B
    Xu, XP
    Antsaklis, PJ
    Michel, AN
    SYSTEMS & CONTROL LETTERS, 1999, 38 (03) : 197 - 207
  • [29] Repetitive learning control for triangular systems with unknown control directions
    Yu, M.
    Ye, X.
    Qi, D.
    IET CONTROL THEORY AND APPLICATIONS, 2011, 5 (17): : 2045 - 2051
  • [30] Optimal Robust Control of Nonlinear Systems with Unknown Dynamics via NN Learning with Relaxed Excitation
    Luo, Rui
    Peng, Zhinan
    Hu, Jiangping
    ENTROPY, 2024, 26 (01)