Neural Stochastic Contraction Metrics for Learning-based Control and Estimation
Name
2011.03168.pdf
Description
Accepted version
Size
946.82 KB
Format
Adobe PDF
Checksum (MD5)
697cc3073c95da9af2cefef18b724a00
Author(s)
Tsukamoto, Hiroyasu
Chung, Soon-Jo
Slotine, Jean-Jacques E
Date Issued
2021
Journal
IEEE Control Systems Letters
Publisher
Institute of Electrical and Electronics Engineers (IEEE)
Citation
Tsukamoto, Hiroyasu, Chung, Soon-Jo and Slotine, Jean-Jacques E. 2021. "Neural Stochastic Contraction Metrics for Learning-based Control and Estimation." IEEE Control Systems Letters, 5 (5).
Version
Author's final manuscript
Abstract
We present Neural Stochastic Contraction Metrics (NSCM), a new design framework for provably-stable learning-based control and estimation for a class of stochastic nonlinear systems. It uses a spectrally-normalized deep neural network to construct a contraction metric and its differential Lyapunov function, sampled via simplified convex optimization in the stochastic setting. Spectral normalization constrains the state-derivatives of the metric to be Lipschitz continuous, thereby ensuring exponential boundedness of the mean squared distance of system trajectories under stochastic disturbances. The trained NSCM model allows autonomous systems to approximate optimal stable control and estimation policies in real-time, and outperforms existing nonlinear control and estimation techniques including the state-dependent Riccati equation, iterative LQR, EKF, and the deterministic NCM, as shown in simulation results.
MIT Department
Massachusetts Institute of Technology. Nonlinear Systems Laboratory
Terms of Use
Creative Commons Attribution-Noncommercial-Share Alike
Persistent DSpace Link
DOI of Published Version
10.1109/LCSYS.2020.3046529
