SFKD combines a fiber-bundle latent manifold, environment-conditioned Koopman operators, and contraction-constrained residuals to certify input-to-state stability while improving path-tracking performance under variable conditions.
Title resolution pending
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
fields
eess.SY 4years
2026 4representative citing papers
The paper introduces ControlSynth Neural ODEs and an ICODE-MPPI controller that certify stability via tractable LMIs and demonstrate improved tracking performance on vehicle and oscillator benchmarks.
Delay-dependent ISS conditions via Lyapunov-Krasovskii LMIs plus a stability-constrained Koopman observer yield 35% lower velocity RMSE and 67% better speed tracking than EKF and FOC on a PMSM drive.
ICNDM jointly learns a controlled neural vector field and a Riemannian metric that enforce input-to-state contraction, cutting long-horizon rollout error on chaotic oscillators and a PMSM drive.
citing papers explorer
-
Stable Fiber-Koopman Residual Dynamics for Environment-Constrained Robust Control
SFKD combines a fiber-bundle latent manifold, environment-conditioned Koopman operators, and contraction-constrained residuals to certify input-to-state stability while improving path-tracking performance under variable conditions.
-
Safe Data-Driven Control and Dynamical Learning via Constrained Neural Architectures and Koopman Operators
The paper introduces ControlSynth Neural ODEs and an ICODE-MPPI controller that certify stability via tractable LMIs and demonstrate improved tracking performance on vehicle and oscillator benchmarks.
-
Stability Analysis and Data-Driven State Estimation for Generalized Persidskii Systems with Time Delays: Theory and Experimental Validation on PMSM Drives
Delay-dependent ISS conditions via Lyapunov-Krasovskii LMIs plus a stability-constrained Koopman observer yield 35% lower velocity RMSE and 67% better speed tracking than EKF and FOC on a PMSM drive.
-
Learning Stable Controlled Dynamical Systems via Input-Contraction Neural Differential Models
ICNDM jointly learns a controlled neural vector field and a Riemannian metric that enforce input-to-state contraction, cutting long-horizon rollout error on chaotic oscillators and a PMSM drive.