arXiv subjects

Kui Xie

Publications and source records attributed to Kui Xie.

3 recordsLinked to original sources

arXiv2026

Learning practically stabilizing output-feedback nonlinear controllers

This paper addresses the problem of learning an output-feedback surrogate controller offline that approximates a given, possibly computationally expensive, nonlinear controller-observer pair. The surrogate is modeled as a recurrent dynamical system and is trained to imitate closed-loop input/output trajectories generated by the given controller. Beyond imitation accuracy, the offline training problem promotes input-to-state practical stability by incorporating estimated state trajectories to learn a candidate Lyapunov function. The approach is validated on a nonlinear continuous stirred tank reactor, where constraint satisfaction and practical stability are assessed through a probabilistic validation approach. The numerical results highlight the benefit of jointly learning the Lyapunov function by comparing against an imitation-only baseline.

eess.SY↗

arXiv2025

Online design of experiments by active learning for nonlinear system identification

We investigate the use of active-learning (AL) strategies to generate the input excitation signal at runtime for system identification of linear and nonlinear autoregressive and state-space models. We adapt various existing AL approaches for static model regression to the dynamic context, coupling them with a Kalman filter to update the model parameters recursively, and also cope with the presence of input and output constraints. We show the increased sample efficiency of the proposed approaches with respect to random excitation on different nonlinear system identification benchmarks.

eess.SY↗

arXiv2025

Active Learning-Based Input Design for Angle-Only Initial Relative Orbit Determination

Accurate relative orbit determination is a significant challenge in modern space operations, particularly when relying only on angular measurements. The inherent observability limitations of this approach make initial state estimation difficult, directly impacting mission safety and performance. This work proposes a hybrid estimation and control strategy for autonomous rendezvous. An active learning (AL) based algorithm designs the initial input control sequence by maximizing the exploration of the output space, thereby enhancing the observability of the initial relative state for the angle-only initial relative orbit determination (IROD) problem. The IROD solution provides a batch estimate of the initial relative state and its analytical covariance, which quantifies the estimation quality and determines the transition point to recursive filtering. Once the uncertainty is sufficiently low, an Extended Kalman Filter (EKF) is initialized with the IROD solution and takes over for sequential estimation, providing state estimates to a Model Predictive Controller (MPC) to complete the rendezvous. The proposed framework is validated through numerical simulations, demonstrating its ability to reliably resolve the scale ambiguity, outperform baseline excitation strategies, and successfully execute an end-to-end rendezvous from initial estimation to final approach.

eess.SY↗