Robust Taylor-Lagrange Control for Safety-Critical Systems with Guaranteed Feasibility
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Robotics Radio. Generated commentary on the latest robotics and control papers.
Rosa: I'm Rosa.
Dev: And I'm Dev, here with Taro.
Rosa: We're looking at the paper: "Robust Taylor-Lagrange Control for Safety-Critical Systems with Guaranteed Feasibility".
Dev: Here's the gist of the paper. Solving safety-critical control problem has widely adopted the Control Barrier Function (CBF) method <ref:2602.20076#pg2>.
Rosa: The recently proposed Taylor-Lagrange Control (TLC) method addresses this limitation, but is vulnerable to the feasibility preservation problem (e.g., inter-sampling effect) <ref:2602.20076#pg2>.
Dev: In this paper, we propose a robust TLC (rTLC) method to address the feasibility preservation problem <ref:2602.20076#pg2>.
Rosa: Pruning redundant reasoning tokens The rTLC method expands the safety function at an order higher than the relative degree of the function using Taylor’s expansion with Lagrange remainder <ref:2602.20076#pg2>.
Title and authors: Rosa: Now, looking at the title and authors, "Robust Taylor-Lagrange Control for Safety-Critical Systems with Guaranteed Feasibility," it tells you they aren't just tweaking an existing tool; they’re proposing a new framework to handle safety in nonlinear systems.
Dev: They mention that standard Control Barrier Functions only give you a sufficient condition for safety, which means the function is non-negative for all times, but not necessarily *for* all times without extra work.
Taro: So this paper is about moving from just knowing you *might* be safe to actually guaranteeing safety across the whole operation using this new Taylor-Lagrange Control method.
Rosa: And they claim this robust Taylor-Lagrange Control method has a major advantage: it naturally handles the feasibility preservation problem, like that inter-sampling effect we talked about earlier.
Dev: They say it does this with only one hyper-parameter, which is just the size of the discretization time interval during implementation, and that's much smaller than what other methods require.
Taro: That makes a difference in terms of practical deployment because we don't need to fine-tune a bunch of different parameters just to keep things stable when the sampling isn't perfect.
The paper's summary: Rosa: So, summarizing what the paper says, they propose expanding the safety function at an order higher than the relative degree of the function using Taylor’s expansion with a Lagrange remainder term.
Dev: This expansion allows them to explicitly show up the control at time t zero instead of some future time in that standard Taylor-Lagrange Control method <ref:2602.20076#pg1>.
Taro: That explicit dependence on t zero is what lets them work with the current state and current control input, which seems to be the main hurdle for real-time implementation <ref:2602.20076#pg1>.
Rosa: They define the robust Taylor-Lagrange Control function by expanding it at an order of m+one instead of just m in Taylor’s theorem, which leads to a remainder term bounded by R min <ref:2602.20076#pg1>.
Dev: The resulting constraint is written as: "sup u(t zero) in U ' mX-one k=zero L k f h(x(t zero)) k! t k + L m f h(x(t zero)) m <ref:2602.20076#pg1>! t m + LgL m-one f h(x(t zero))u t zero m <ref:2602.20076#pg1>! t m + R min zero <ref:2602.20076#pg1>."
Taro: That equation shows that the safety condition depends on the current control u(t zero) and terms involving the future time xi, but because of this expansion, it's tied more closely to the current time <ref:2602.20076#pg1>.
The paper's improvements: Rosa: One big improvement they highlight is how this method addresses inter-sampling effect without needing extra considerations, which is a huge win for practical application.
Dev: They show in Theorem two that any controller u(t zero) that satisfies the robust Taylor-Lagrange Control constraint in equation (sixteen) ensures the safety function stays non-negative for all future times t' between t zero and t.
Taro: That property means the control you calculate right now, at state x(t zero), automatically handles the issue of sampling irregularly, so it doesn't need complex event-triggered logic layered on top <ref:2602.20076#pg1>.
Rosa: They also show that this method only needs one hyper-parameter: the discretization time interval size during implementation, which is much less than what you see in other methods like event-triggered approaches.
Dev: That makes the whole setup way simpler for control engineers because it reduces the tuning burden significantly while still guaranteeing safety.
Conclusion: Rosa: So, to wrap up on "Robust Taylor-Lagrange Control for Safety-Critical Systems with Guaranteed Feasibility," they’ve proposed this robust method using a higher-order expansion of the safety function via Taylor’s theorem and a Lagrange remainder term.
Dev: The main point is that it solves the feasibility preservation problem, like inter-sampling effect, by making the control show up at the current time t zero instead of some future time <ref:2602.20076#pg1,show up at the current time>.
Taro: It means we can define a state feedback controller K rTLC(x(t zero)) that satisfies the safety constraint without needing to look at an unknown point xi in the interval (t zero t) <ref:2602.20076#pg1>.
Rosa: They verified its effectiveness using an adaptive cruise control problem, showing it performs well and compares favorably against other methods like event-triggered TLC.
Dev: The limitation they state is that while they address inter-sampling effect with one hyper-parameter, the method still involves a specific discretization time interval size that needs to be chosen carefully for the system's dynamics.
Taro: They also plan to study the tradeoff between how conservative this robust Taylor-Lagrange control is and whether it maintains feasibility in different scenarios.
Rosa: Overall, this work provides a way to ensure safety in nonlinear systems by using this higher-order expansion, naturally handling inter-sampling effect with minimal tuning.
Dev: It's a solid step toward making safety guarantees more practical for real-time control loops without adding a huge number of parameters.
Wei Xiao, Christos Cassandras, Anni Li
Worcester Polytechnic Institute · MIT CSAIL · The University of North Carolina at Charlotte · Boston University
eess.SY, cs.AI, cs.RO, cs.SY
Submitted: 2026-02-23
Updated: 2026-10-29
Comments: This version addressed one technical problem of the inter-sampling effect in rTLC
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Importance score: 76/100
The gist: Solving safety-critical control problem has widely adopted the Control Barrier Function (CBF) method <ref:2602.20076#pg2>.
Terminology
Summary
Solving safety-critical control problem has widely adopted the Control Barrier Function (CBF) method <ref:2602. p. 020076#pg2>. The recently proposed Taylor-Lagrange Control (TLC) method addresses this limitation, but is vulnerable to the feasibility preservation problem (e. p. 0g., inter-sampling effect) <ref:2602. p. 020076#pg2>. In this paper, we propose a robust TLC (rTLC) method to address the feasibility preservation problem <ref:2602. p. 020076#pg2>.
Pruning redundant reasoning tokens
The rTLC method expands the safety function at an order higher than the relative degree of the function using Taylor’s expansion with Lagrange remainder <ref:2602.20076#pg2>. This expansion allows the control to explicitly show up at the current time instead of the future time in the TLC method <ref:2602. p. 020076#pg2>. The rTLC method naturally addresses the feasibility preservation problem with only one hyper-parameter (the discretization time interval size during implementation), which is much less than its counterparts <ref:2602. p. 020076#pg2>.
How the training signal is built
The robust Taylor-Lagrange Control (rTLC) function is defined by expanding the function at an order of m + 1 instead of m in the Taylor’s theorem <ref:2602.20076#pg4>. This leads to a remainder term R(x(ξ),u(ξ), u˙(ξ)) which is bounded by Rmin <ref:2602. p. 020076#pg5>. The robust TLC function is defined as h: R n → R if sup u(t0)∈U "mX−1 k=0 L k fh(x(t0)) k! ∆t k + L m f h(x(t0)) m! ∆t m + LgL m−1 f h(x(t0))u t0 m! p. 1 ∆t m + Rmin >= 0. p. 2
Where the gains come from
The robustness of the rTLC method stems from its ability to address the inter-sampling effect without additional considerations. The proof of Theorem 2 shows that any controller u(t0) that satisfies the rTLC constraint in (16) ensures h(x(t′)) >= 0 for all t′ ∈ [t0, t]. p. 3 This property allows the controller u(t0) obtained at the current state x(t0) to naturally address the intersampling effect. p. 3
Case Studies and Results
The effectiveness of the proposed rTLC method is verified on an adaptive cruise control problem. The comparison between methods in Table II shows that while the time-driven TLC may violate the safety constraint due to the inter-sampling effect, the rTLC may be conservative as shown with a larger h(x) steady value for rTLC in the last framework of Fig. p. 5 1. Decreasing ∆t could significantly reduce this conservativeness. p. 5
Conclusion
We proposed a robust Taylor-Lagrange control method for ensuring the safety of nonlinear control systems using a higher-order expansion of the safety function with Taylor’s theorem with Lagrange remainder. The proposed robust Taylor-Lagrange control method can naturally address the inter-sampling effect while introducing much less hyperparameters than other methods. p. 5 We applied the proposed method to an adaptive cruise control problem to demonstrate the effectiveness of our methods. p. 5 In the future, we plan to study the tradeoff between conservativeness and feasibility of the proposed robust Taylor-Lagrange control method. p. 5 We will also investigate how to compute a tighter reachable set for the Lagrange remainder to reduce the conservativeness of the proposed robust Taylor-Lagrange control method. p. 5
REFERENCES
arXiv:2602.20076v1 arXiv preprint arXiv:2512. p. 6
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Robust Safety Guarantee for Control Systems: The improved system can guarantee forward invariance of a safe set C for nonlinear systems by using Taylor-Lagrange Control (TLC) or robust TLC (rTLC). This addresses the limitation that standard Control Barrier Functions only provide a sufficient condition for safety, ensuring the safety function remains non-negative for all times.
-
Addressing Inter-sampling Effect in Safety Constraints: The system can naturally address the inter-sampling effect without introducing complex hyper-parameters by using rTLC. The paper states that
the proposed rTLC method naturally addresses the feasibility preservation problem with only one hyperparameter (the discretization time interval size during implementation), which is much less than its counterparts.
-
Real-Time Control Policy Generation: The system can define a state feedback controller Krtlc(x(t0)) that satisfies the safety constraint by making
no dependence on x(ξ)
in the constraint definition. This allows for direct calculation of the required control u at the current time t0, facilitating practical implementation. -
Optimal Control Synthesis with Safety Constraints: The system can solve constrained optimal control problems, such as minimizing energy consumption while maintaining a safety constraint h(x) >= 0, by reformulating them into a sequence of Quadratic Programs (QPs). This is achieved by using rTLC to enforce the safety constraint and Taylor-Lagrange Stability (TLS) functions for state convergence.
-
Application in Adaptive Cruise Control: The improved system can implement adaptive cruise control with guaranteed safety for vehicle dynamics. By applying the proposed rTLC method, the system ensures that
the safety of the system can be guaranteed
while comparing its performance against other methods like HOCBF and event-triggered TLC.
Sources
Related papers
- Industrial Load Modeling and Optimization for Market-Based Interaction with Power Systems
- Exact and Approximate Convex Reformulation of Linear Stochastic Optimal Control with Chance Constraints
- Certified Set Convergence for Piecewise Affine Systems
- Symbolic Control for Stochastic Systems via Finite Parity Games
- Multiobjective optimization-based design and dispatch of islanded, hybrid microgrids for remote, off-grid communities in sub-Saharan Africa
- Physiology-informed layered sensing for intelligent human-exoskeleton interaction