Neural Stochastic Contraction Metrics for Learning-based Control and Estimation

Abstract

We present Neural Stochastic Contraction Metrics (NSCM), a new design framework for provably-stable robust control and estimation for a class of stochastic nonlinear systems. It uses a spectrally-normalized deep neural network to construct a contraction metric, 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 NSCM framework allows autonomous agents 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 neural contraction metric, as illustrated in simulation results.

arXiv commentIEEE CONTROL SYSTEMS LETTERS (L-CSS), preprint version, accepted Dec. 2020 (DOI: 10.1109/LCSYS.2020.3046529). https://ieeexplore.ieee.org/document/9302618

References in corpus (4)

• Neural Lander: Stable Drone Landing Control using Learned Dynamics
• Neural Contraction Metrics for Robust Estimation and Control: A Convex Optimization Approach
• Robust Controller Design for Stochastic Nonlinear Systems via Convex Optimization
• A Comparison of LPV Gain Scheduling and Control Contraction Metrics for Nonlinear Control

Cited by in corpus (3)

• Contraction Theory for Nonlinear Stability Analysis and Learning-based Control: A Tutorial Overview
• Learning-based Robust Motion Planning with Guaranteed Stability: A Contraction Theory Approach
• $k$-Contraction in a Generalized Lurie System