Internal-Time Gating for Uncertainty-Aware Switching Control

This work presents an internal-time–based gating mechanism for uncertainty-aware switching control systems. The proposed framework addresses stability and safety challenges in learning-enabled and adaptive controllers by regulating the frequency of high-impact control actions, rather than enforcing per-step admissibility constraints. An internal-time variable is introduced whose evolution is modulated by a predictive uncertainty measure. Access to high-impact (potentially non-contractive) control modes is permitted only when the accumulated internal time exceeds a prescribed threshold and uncertainty is sufficiently low. This mechanism induces an endogenous switching signal between safe and high-impact dynamics. The closed-loop system is modeled as a two-mode switched system, and stability is analyzed using Lyapunov methods and average dwell-time theory. Explicit conditions are derived that guarantee uniform ultimate boundedness and asymptotic stability in the disturbance-free case. Constructive design and calibration rules are provided to estimate contraction and expansion bounds from short-horizon data and translate them into explicit parameters for the internal-time threshold. The framework is fully self-contained from a control-theoretic perspective and does not rely on any physical interpretation of time or entropy beyond their operational use as uncertainty proxies. It complements existing Lyapunov- and barrier-based safety approaches by shifting part of the safety burden from instantaneous action filtering to temporal regulation of mode switching. The proposed method is applicable to a broad class of learning-based, adaptive, and hybrid control systems. Relation to Prior Work This work builds conceptually on the author’s foundational framework on internal-time and endogenous governance mechanisms, available as a separate Zenodo record. The present contribution is technically independent and focuses exclusively on control-theoretic formulation, stability analysis, and constructive design rules. Notes This record provides a standalone technical contribution with an independent DOI. The associated foundational work is cited for conceptual motivation only and is not required to understand or apply the results presented here.