Learning Certified Neural Network Controllers Using Contraction and Interval Analysis
IEEE Control Systems Letters · 2026 · 10 · 1615-1620
DOI: 10.1109/lcsys.2026.3707155 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Control Systems Letters, 2026. Verfasst von Akash Harapanahalli und 2 weiteren Autor:innen. Die Arbeit nennt 22 Literaturangaben. Gefördert von: National Science Foundation, Air Force Office of Scientific Research.
Abstract (Originalsprache)
We present a novel framework that jointly trains a neural network controller and a neural Riemannian metric with rigorous closed-loop contraction guarantees using formal bound propagation. Directly bounding the symmetric Riemannian contraction linear matrix inequality causes unnecessary overconservativeness due to poor dependency management. Instead, we analyze an asymmetric matrix function $G$ , where $2^{n}$ GPU-parallelized corner checks of its interval hull verify that an entire interval subset $X$ is a contraction region in a single shot. This eliminates the sample complexity problems encountered with previous Lipschitz-based guarantees. Additionally, for control-affine systems under a Killing field assumption, our method produces an explicit tracking controller capable of exponentially stabilizing any dynamically feasible trajectory using just two forward inferences of the learned policy. Using JAX and immrax for linear bound propagation, we apply this approach to a full 10-state quadrotor model. In $\pi $ , a neural contraction metric $\Theta $ , and a verified 10-dimensional contraction region $X$ .
Förderung
- National Science Foundation
- Air Force Office of Scientific Research
Zitieren
APA
Harapanahalli, A., Coogan, S., & Davydov, A. (2026). Learning Certified Neural Network Controllers Using Contraction and Interval Analysis. IEEE Control Systems Letters, 10, 1615-1620. https://doi.org/10.1109/lcsys.2026.3707155
Vancouver
Harapanahalli A, Coogan S, Davydov A. Learning Certified Neural Network Controllers Using Contraction and Interval Analysis. IEEE Control Systems Letters. 2026;10:1615-1620. doi:10.1109/lcsys.2026.3707155
BibTeX
@article{Harapanahalli2026,
author = {Harapanahalli, Akash and Coogan, Samuel and Davydov, Alexander},
title = {Learning Certified Neural Network Controllers Using Contraction and Interval Analysis},
journal = {IEEE Control Systems Letters},
year = {2026},
volume = {10},
pages = {1615--1620},
doi = {10.1109/lcsys.2026.3707155}
}
EUVIMED