Integral Control Barrier Functions with Input Delay: Prediction, Feasibility, and Robustness

Authors: Adam K. Kiss, Ersin Das, Tamas G. Molnar, Aaron D. Ames Date: 2026-03-25 Paper ID: arxiv:2603.24566

Summary

This paper presents a safety-critical control framework for nonlinear systems that must operate subject to state and input constraints while suffering from an input time delay. The core contribution is the integration of predictor feedback within dynamically defined Integral Control Barrier Functions (ICBFs) to compensate for the inherent delay in the system dynamics. The authors formulate the safe control design as a Quadratic Program (QP) and derive a closed-form feasibility condition that guarantees forward invariance under delay when both state and input constraints are active. Furthermore, the framework is extended using tunable robust ICBFs to maintain safety guarantees despite uncertainties in the delay estimation. The proposed method is validated on an adaptive cruise control system exhibiting actuation delay.

Key Contributions

• Development of a safety-critical control framework for nonlinear systems with input delay using dynamically defined integral controllers.
• Integration of predictor feedback into the dynamically defined control law to explicitly compensate for the effect of input delays.
• Derivation of a closed-form feasibility condition for the resulting quadratic program controller when simultaneous state and input constraints are active.
• Introduction of tunable robust Integral Control Barrier Functions (ICBFs) to ensure robustness against prediction errors arising from delay uncertainty.

Limitations

The current work focuses on systems where the delay is known or estimated sufficiently well for the predictor to be effective, and robustness analysis might be limited to specific classes of prediction errors.

Open Questions & Future Work

• icbf-two-constraint-compatibility-robust
• robust-icbf-dynamics-uncertainty
• input-delay-safety-without-initial-history-assumption
• closed-form-state-only-icbf-feasibility
• experimental-validation-embedded-platforms

Limitations

The current work focuses on systems where the delay is known or estimated sufficiently well for the predictor to be effective, and robustness analysis might be limited to specific classes of prediction errors.

Links

• ArXiv Abstract
• PDF