Sampling-Horizon Neural Operator Predictors for Nonlinear Control Under Delayed Inputs
Abstract
Control systems often operate with input delays and sampled measurements. However, practical implementations require solving an implicit ODE online, resulting in intractable computational cost. Moreover, predictor formulations typically assume continuously available state measurements, whereas in practice measurements may be sampled, irregular, or temporarily missing due to hardware faults. In this work, we develop two neural-operator predictor-feedback designs for nonlinear systems with delayed inputs and sampled measurements. In the first design, we introduce a sampling-horizon prediction operator that maps the current measurement and input history to the predicted state trajectory over the next sampling interval. In the second design, the neural operator approximates only the delay-compensating predictor, which is then composed with the closed-loop flow between measurements. The first approach requires uniform sampling but yields residual bounds that scale directly with the operator approximation error. In contrast, the second accommodates non-uniform, but bounded sampling schedules at the cost of amplified approximation error, revealing a practical tradeoff between sampling flexibility and approximation sensitivity for the control engineer. For both schemes, we establish semi-global practical stability with explicit neural operator error-dependent bounds. Numerical experiments on a 6-link nonlinear robotic manipulator demonstrate accurate tracking and substantial computational speedup of 25× over a baseline approach.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
Cites
Learning nonlinear operators via DeepONet based on the universal approximation theorem of operators
Lu, L., Jin, P., Pang, G., Zhang, Z., & Karniadakis, G. E. (2021) · Nature Machine Intelligence · doi:10.1038/s42256-021-00302-5 · Cited by 3,656 · see citing works
Smooth stabilization implies coprime factorization
Sontag, E. D. (1989) · IEEE Transactions on Automatic Control · doi:10.1109/9.28018 · Cited by 2,748 · see citing works
On characterizations of the input-to-state stability property
Sontag, E. D., & Wang, Y. (1995) · Systems & Control Letters · doi:10.1016/0167-6911(94)00050-6 · Cited by 1,815 · see citing works
Introduction to Time-Delay Systems
Fridman, E. (2014) · Systems & control · doi:10.1007/978-3-319-09393-2 · Cited by 1,123 · see citing works
Delay Compensation for Nonlinear, Adaptive, and PDE Systems
Krstić, M. (2009) · Systems & control · doi:10.1007/978-0-8176-4877-0 · Cited by 1,064 · see citing works
Input Delay Compensation for Forward Complete and Strict-Feedforward Nonlinear Systems
Krstić, M. (2009) · IEEE Transactions on Automatic Control · doi:10.1109/tac.2009.2034923 · Cited by 393 · see citing works
Neural Networks-Based Adaptive Control for Nonlinear State Constrained Systems With Input Delay
Li, D.-P., Liu, Y.-J., Tong, S., Chen, C. L. P., & Li, D.-J. (2019) · IEEE Transactions on Cybernetics · doi:10.1109/tcyb.2018.2799683 · Cited by 382 · see citing works
Adaptive neural network control of nonlinear systems with unknown time delays
Ge, S. S., Hong, F., & Lee, T. H. (2003) · IEEE Transactions on Automatic Control · doi:10.1109/tac.2003.819287 · Cited by 348 · see citing works
Nonlinear Control Under Nonconstant Delays
Bekiaris-Liberis, N., & Krstic, M. (2013) · doi:10.1137/1.9781611972856 · Cited by 216 · see citing works
Self-citationTruncated Predictor Feedback for Time-Delay Systems
Zhou, B. (2014) · Springer eBooks · doi:10.1007/978-3-642-54206-0 · Cited by 204 · see citing works
Predictor Feedback for Delay Systems: Implementations and Approximations
Karafyllis, I., & Krstic, M. (2017) · Systems & Control: Foundations & Applications · doi:10.1007/978-3-319-42378-4 · Cited by 135 · see citing works
Adaptive neural network predictive control for nonlinear pure feedback systems with input delay
Na, J., Ren, X., Shang, C., & Guo, Y. (2012) · Journal of Process Control · doi:10.1016/j.jprocont.2011.09.003 · Cited by 101 · cited 1× · see citing works
“However, there are a series of works with finite-dimensional neural network (NN) approximations worth recognizing, including adaptive NN control under unknown delays [1], predictive control for pure-feedback systems with input delay [2], and state-constrained control with input delay [3].”
Neural Operators for Bypassing Gain and Control Computations in PDE Backstepping
Bhan, L., Shi, Y., & Krstic, M. (2024) · IEEE Transactions on Automatic Control · doi:10.1109/tac.2023.3347499 · Cited by 58 · see citing works
Self-citationNumerical schemes for nonlinear predictor feedback
Karafyllis, I., & Krstić, M. (2014) · Mathematics of Control Signals and Systems · doi:10.1007/s00498-014-0127-9 · Cited by 42 · see citing works
Nonlocality and Nonlinearity Implies Universality in Operator Learning
Lanthaler, S., Li, Z., & Stuart, A. M. (2025) · Constructive Approximation · doi:10.1007/s00365-025-09718-3 · Cited by 25 · see citing works
Nonlocality and Nonlinearity Implies Universality in Operator Learning
Lanthaler, S., Li, Z., & Stuart, A. M. (2023) · arXiv · doi:10.48550/arxiv.2304.13221 · Cited by 25 · see citing works
Deep Operator Neural Network Model Predictive Control
de Jong, T., Shukla, K., & Lazar, M. (2025) · IEEE Open Journal of Control Systems · doi:10.1109/ojcsys.2025.3614875 · Cited by 6 · see citing works
Neural Operators for Predictor Feedback Control of Nonlinear Delay Systems
Bhan, L., Qin, P., Krstic, M., & Shi, Y. (2024) · arXiv · doi:10.48550/arxiv.2411.18964 · Cited by 1 · see citing works
Self-citationBhan, L., Krstic, M., & Shi, Y. (2025) · arXiv · doi:10.48550/arxiv.2509.26443 · cited 2×
“More recently, neural operator–based predictors have been proposed as a data-driven alternative that learns the predictor mapping directly from trajectory data [9], [10], [11], [12].”
Self-citation
Flagged first, then by citations
Research works printed but not linked: 6
- Bhan, Neural operators for predictor feedback control of nonlinear delay systems, 179
- Bajraktari, Delay compensation of multi-input distinct delay nonlinear systems via neural operators, 1
- Li, Fourier neural operator for parametric partial differential equations, 1
- Bhan, 2026, Neural Operator Predictors For Sampled Measurements
- Spong, 2006, Robot Modeling and Control
- Krstic, 1995, Nonlinear Adapt. Control Design
Cited by
Still being checked.
- Authors
- Bhan, L., Quawas, P., Krstic, M., & Shi, Y.
- Affiliations
- University of California San Diego (US)
- University of California, San Diego, CA, USA
- Venue
- IEEE Control Systems Letters
- Published
- 2026
- Volume
- 10
- Pages
- 475-480
- ISSN
- 2475-1456 (ISSN-L)
- OpenAlex
- W7163035394
- Cited by
- 0
- References
- 19
- FWCI
- 0.00
- Type
- journal article
- Language
- en
- Open access
- closed
- Licence
- Locations
- IEEE Control Systems Letters (publishedVersion)
- Funders
- Department of Energy (DESC0024386)
- Air Force Office for Scientific Research (FA9550-23-1-0535)
- National Science Foundation (NSF) under Grant (ECCS-2151525)
- DOE (DE-SC0025495)
- Schmidt Sciences AI2050 Early Career Fellowship
- NSF (2442689)
- Topics
- Physical Sciences › Engineering › Control and Systems Engineering › Advanced Control Systems Optimization
- Physical Sciences › Engineering › Control and Systems Engineering › Stability and Control of Uncertain Systems
- Physical Sciences › Engineering › Control and Systems Engineering › Adaptive Control of Nonlinear Systems
- Keywords
- neural operators, predictor feedback, input delay compensation, sampled measurements, nonlinear control, semi-global practical stability, residual bounds, approximation error, non-uniform sampling, robot manipulator control
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
