Verification and Forward Invariance of Control Barrier Functions for Differential-Algebraic Systems

arXiv:2603.13509 · eess.SY, cs.RO, cs.SY · Submitted 2026-03-13 · Read on arXiv

Listen

Transcript

Introduction to the show: ident: Robotics Radio. Generated commentary on the latest robotics and control papers.

Rosa: I'm Rosa, and with me are Dev and Taro, guest researcher.

Dev: Today's paper: "Verification and Forward Invariance of Control Barrier Functions for Differential-Algebraic Systems".

Rosa: The gist Differential-algebraic equations (DAEs) arise in power networks, chemical processes, and multibody systems, where algebraic constraints encode physical conservation laws.

Dev: First, who's behind it and why it matters.

Paper summary: Rosa: So to go into more detail about this paper, "Verification and Forward Invariance of Control Barrier Functions for Differential-Algebraic Systems", the core thesis is that existing CBF methods are not directly applicable to DAEs because of the conflicts between the safety condition and the constraint manifold.

Dev: The paper solves this by introducing DAE-aware CBFs which incorporate that differential-algebraic structure through projected vector fields. They derive conditions that guarantee forward invariance of safe sets while preserving those algebraic constraints and they extend this approach to higher index DAEs.

Taro: The real issue they're tackling is how the algebraic constraints restrict the allowable state trajectories, and standard CBFs fail because of that restriction on where you can move.

Rosa: They show that by using projected vector fields, you can make compatibility conditions explicit, including exactly how those algebraic state derivatives are determined. This involves decomposing the Jacobian of the constraint with respect to dynamic variables and algebraic variables to derive a specific compatibility condition.

Dev: That specific condition is written as "Jd(x)fd(x) + gd(x)u + Ja(x) ˙xa = zero", and Lemma one connects the differentiation index nu to the relative degree d by saying d equals nu plus d prime minus one <ref:2603.13509#pg1>.

Taro: That relationship between the index nu and the relative degree d is a key piece of information because it tells you when those CBF constraints stay feasible even when there are hidden algebraic constraints involved.

Rosa: They then establish sufficient conditions for positive invariance of the safe set D intersection M through Theorem four and Theorem five depending on whether it's an index-one system or a higher index system.

Dev: For index-one systems, safety requires satisfying two things: first, a condition for manifold invariance involving P(x)Jd(x)fd(x) plus eta one of x gd(x)u equals zero, and second, a barrier nonnegativity condition Lf hat b of x plus Lg hat b of x u is greater than or equal to negative alpha naught zero of b <ref:2603.13509#pg1>.

Taro: For those higher-index systems, they use Corollary four point one which extends this to a hierarchy involving the High-Order CBF framework, where invariance demands satisfying conditions twenty for every k from one up to nu.

Rosa: Beyond just the safety conditions, they build a systematic verification framework that checks for geometric correctness and feasibility of these DAE-aware CBFs. They use sum-of-squares certificates for polynomial systems and satisfiability modulo theories for nonpolynomial or neural network candidates.

Dev: The final result is Theorem seven which states that if correctness, interior feasibility, and boundary tangency hold, then b of x is a valid DAE-aware CBF.

Taro: It’s important to remember that the paper focuses on verifying the existence of these filters rather than providing a direct method for synthesizing them from safety specifications in real time.

Rosa: That's right, they don't address how you actually synthesize a DAE-aware CBF based purely on what you want the system to do. That is one limitation they point out at the end of the paper.

Conclusion: Rosa: So wrapping up with the conclusion for this work, "Verification and Forward Invariance of Control Barrier Functions for Differential-Algebraic Systems", the authors are focused on proving that their new DAE-aware CBFs actually work under those tricky algebraic constraints.

Dev: The implication is that you can design safety filters for complex systems like power networks or multibody systems even when they have those hidden constraints because this method respects them.

Taro: For someone who just listens to the show, it means we can trust that a control policy designed using this framework will maintain safety without violating the physical laws encoded in those algebraic constraints.

Rosa: Exactly, it’s about giving us a rigorous way to confirm that the filter you're designing won't lead the system into an unsafe region defined by those constraints.

Dev: They did this by tackling systems with index-one and index-two DAEs, showing their framework holds up across different levels of complexity <ref:2603.13509#pg1>.

Taro: It shows that even when the constraints are hidden in a way that isn't immediately obvious from the constraint equations alone, you can still analyze the system's behavior on that manifold.

Rosa: The work moves safety filtering for these systems from just diagnostic tools to something where we can actually verify the constructive methods, even if they aren't synthesizing them automatically.

Dev: So ultimately, this paper provides a solid set of conditions—the necessary and sufficient ones—for what constitutes a valid DAE-aware CBF in terms of geometric correctness and feasibility.

Taro: It sets a foundation for how we can formally analyze control safety in these types of dynamic systems by accounting for the underlying differential-algebraic structure explicitly.

Vanderbilt University

eess.SY, cs.RO, cs.SY

Submitted: 2026-03-13

Updated: 2026-10-13

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 73/100

The gist: The gist Differential-algebraic equations (DAEs) arise in power networks, chemical processes, and multibody systems, where algebraic constraints encode physical conservation laws.

Key concepts

Differential-Algebraic Equations (DAEs): These equations describe systems like power grids or chemical processes where some variables change over time (differential) and others are fixed by algebraic laws, such as conservation of mass. They are complex because the algebraic constraints restrict the possible paths a system can take in its state space.
Projected Vector Fields: These mathematical tools encode the DAE's structure directly into a new set of equations that describe motion only on the allowed constraint manifold (M). By using these fields, researchers can explicitly show how algebraic variables are determined by dynamic ones, making compatibility conditions clear.
Sum-of-Squares (SOS) Certificates: These are mathematical tools used to prove that a polynomial function is always non-negative over a specific region. In this paper, SOS certificates verify the geometric correctness of the DAE-aware CBF design by proving that certain safety conditions hold true for all valid system states.
High-Order CBF (HOCBF) Framework: This extension allows the safety framework to work effectively even when dealing with higher-index DAEs, which have more complex algebraic dependencies. The HOCBF hierarchy systematically checks invariance and safety conditions across multiple levels of constraints, ensuring robust control for more intricate physical systems.

Terminology

Verification — 2× in this paper · explained in 2 episodes
Conservation Laws — 1× in this paper · explained in 3 episodes
Control Barrier Functions (CBFs) — 1× in this paper · explained in 6 episodes

Summary

The gist Differential-algebraic equations (DAEs) arise in power networks, chemical processes, and multibody systems, where algebraic constraints encode physical conservation laws.

Problem Formulation

The core problem addressed is designing safe control policies for DAE systems by constructing DAE-aware CBFs that explicitly account for the differential-algebraic structure through projected vector fields. A fundamental challenge is that standard CBF methods fail because the algebraic constraints restrict allowable state trajectories, leading to potential infeasibility of the CBF-QP when applied to DAEs. The goal is formalized as Problem 1: Given a DAE system described by (1) and a safe set C, design a safe control policy µ: Rnx → Rnu such that the DAE system (1) maintains positive invariance of D∩M for all initial states x(0) ∈ D∩M.

DAE-Aware CBF Construction

The paper introduces a novel framework where projected vector fields encode the differential-algebraic structure and make the role of compatibility conditions explicit, including how algebraic state derivatives are determined. This involves decomposing the Jacobian of the constraint ϕ with respect to dynamic variables xd and algebraic variables xa to derive a compatibility condition (7): Jd(x)fd(x) + gd(x)u + Ja(x) ˙xa = 0. The key result is Lemma 1, which connects the differentiation index ν to the relative degree d: d = ν + d' − 1. This relationship is crucial for understanding when CBF constraints remain feasible in the presence of hidden algebraic constraints.

Safety and Invariance Conditions

The framework establishes sufficient conditions for positive invariance of the safe set D∩M through Theorem 4 and Theorem 5. For index-1 systems, the safety requires satisfying two conditions: P(x)Jd(x)fd(x) + η(1)(x)gd(x)u = 0 for manifold invariance, and Lfˆb(x) + Lgˆb(x)u ≥ −α00(b) for barrier nonnegativity. For higher-index systems, Corollary 4.1 extends this to a hierarchy involving the High-Order CBF (HOCBF) framework, where invariance requires satisfying conditions (20) for all k ∈ 1,..., ν.

Verification Framework

A systematic verification framework is developed to certify geometric correctness and feasibility of candidate DAE-aware CBFs. For polynomial systems, this involves sum-of-squares certificates are provided via SOS programming (28). For nonpolynomial and neural network candidates, the approach uses satisfiability modulo theories for falsification. The final result is Theorem 7, which states that if Correctness, Interior Feasibility, and Boundary Tangency hold ⇒ b(x) is a valid DAE-aware CBF.

Validation and Conclusion

The proposed framework was validated on two systems: a wind turbine system and a flexiblelink manipulator, demonstrating that DAE-aware CBFs maintain invariance of the constraint manifold and ensure feasibility of the safety filter. The experiments confirmed that the DAE-aware CBF-QP controller successfully enforces safety while respecting algebraic constraints for both index-1 and index-2 systems. The framework is shown to extend to higher-index DAEs through highorder projected vector fields that account for hidden constraints and the system’s differentiation index, giving rise to HOCBFs. The manuscript notes that the problem of synthesizing DAE-aware CBFs from safety specifications is not addressed.

--- Page 1 ---

The gist Differential-algebraic equations (DAEs) arise in power networks, chemical processes, and multibody systems, where algebraic constraints encode physical conservation laws.

How it works

The core mechanism involves constructing projected vector fields that encode the differential-algebraic structure and make the role of compatibility conditions explicit. This process transforms the DAE into an equivalent ODE on the constraint manifold M. For index-1 systems, this leads to a projected control-affine representation x˙ = ˆf(x) + ˆg(x)u on M.

Safety and Invariance Conditions

The safety of the set D∩M is ensured by combining manifold invariance with the standard CBF condition. This requires satisfying compatibility condition P(x)Jd(x)fd(x) + η(1)(x)gd(x)u = 0 and the barrier inequality Lfˆb(x) + Lgˆb(x)u ≥ −α00(b). For higher-index systems, Corollary 4.1 requires satisfying conditions (24a) for manifold invariance and (24b) for safety via the HOCBF hierarchy.

Verification Framework

The verification process is structured into two main branches: Correctness and Feasibility. Correctness is checked by searching for SOS polynomials satisfying the Positivstellensatz condition (27). Feasibility is verified using Farkas duality to check the existence of an admissible control input u.

Validation and Conclusion

The framework was validated on two systems, demonstrating that DAE-aware CBFs can maintain safety without violating algebraic constraints. The experiments confirmed that the DAE-aware CBF-QP controller successfully enforces safety while respecting algebraic constraints for both index-1 and index-2 systems. The manuscript notes that the problem of synthesizing DAE-aware CBFs from safety specifications is not addressed.

How it works

Improvements for AI systems

Bold header: DAE-aware CBF implementation in power networks

This improved system can ensure forward invariance of safe sets while remaining consistent with the algebraic constraints by incorporating projected vector fields that encode the differential-algebraic structure. Specifically, it can guarantee that a control input satisfying the barrier condition does not violate the constraint-manifold compatibility condition, which prevents safety violations observed in standard CBF approaches.

Bold header: Safety certification for high-index flexible manipulators

The system can now verify if a candidate barrier function is valid by checking geometric correctness and feasibility using SOS programming for polynomials or SMT-based falsification for neural networks, as described in Theorem 7. This allows formal safety guarantees even for complex, higher-index systems like the index-2 flexible-link manipulator mentioned in the experiments.

Bold header: Adaptive control synthesis respecting algebraic constraints

The system can synthesize a controller that enforces compatibility condition (13a) together with the barrier inequality (13b) by solving a unified affine system problem, which is equivalent to checking dual feasibility conditions (Proposition 2). This ensures that the resulting control input respects both manifold invariance and positive invariance simultaneously.

Abstract

Differential-algebraic equations (DAEs) arise in power networks, chemical processes, and multibody systems, where algebraic constraints encode physical conservation laws. The safety of such systems is critical, yet safe control is challenging because algebraic constraints restrict allowable state trajectories. Control barrier functions (CBFs) provide computationally efficient safety filters for ordinary differential equation (ODE) systems. However, existing CBF methods are not directly applicable to DAEs due to potential conflicts between the CBF condition and the constraint manifold. This paper introduces DAE-aware CBFs that incorporate the differential-algebraic structure through projected vector fields. We derive conditions that ensure forward invariance of safe sets while preserving algebraic constraints and extend the framework to higher-index DAEs. A systematic verification framework is developed, establishing necessary and sufficient conditions for geometric correctness and feasibility of DAE-aware CBFs. For polynomial systems, sum-of-squares certificates are provided, while for nonpolynomial and neural network candidates, satisfiability modulo theories are used for falsification. The approach is validated on wind turbine and flexible-link manipulator systems.