Zonotope-Based Elastic Tube Model Predictive Control

arXiv:2509.19824 · eess.SY, cs.SY · Submitted 2025-09-24 · Read on arXiv

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: "Zonotope-Based Elastic Tube Model Predictive Control".

Dev: Here's the gist of the paper. Tube-based Model Predictive Control (MPC) is a robust control framework for constrained linear systems under additive disturbance, focusing on reducing numerical complexity through scaled zonotopes The paper proposes a novel class of scaled-zonotope inclusion conditions that alleviates the need for a priori specification of certain set-containment constraints and achieves significant reductions in complexity Problem Formulation and System Modeling The work centers on extending elastic tube MPC by introducing a novel linear encoding for zonotope containment that generalizes standard conditions, thereby circumventing the need to preselect auxiliary variables This formulation extends naturally to scaled zonotopes, eliminating the bilinearities typically associated with scaling factors and avoiding suboptimal parameter selection The core optimization problem is reformulated to allow for an explicit quantification of computational trade-offs, contrasting with polyhedral scaling methods that involve bilinear terms Scaled Zonotopic Inclusion Conditions The paper introduces a sufficient condition for the inclusion of zonotopic sets, which is given by Lemma 1, stating that the existence of matrices Γi and a vector γ verifying certain conditions ensures the inclusion Corollary 1 provides an equivalent condition for scaled zonotopes involving a coefficient matrix Φ, which is linear in both auxiliary variables and scaling factors, facilitating its integration into large-scale optimization frameworks Computation and Complexity Analysis A comprehensive complexity analysis is provided for both the polyhedral and the zonotopic setting, illustrating the trade-off between an enlarged domain of attraction and the required computational effort Table 1 summarizes various scenarios for this purpose, indicating that zonotopic variants outperform the polyhedral case in terms of computational time for certain system models The complexity is measured by the number of decision variables and constraints, showing that the 'with centers' case introduces an additional overhead Recursive Feasibility and Stability Guarantees Necessary and sufficient conditions for zonotopic robust control tubes were recently established using vertex control, but this method is computationally prohibitive due to combinatorial vertex growth The proposed method leverages scalable linear encodings that circumvent vertex enumeration entirely, prioritizing numerical tractability The final formulation verifies admissibility conditions (4)–(6) and recursive feasibility guarantees (26)–(28) through the set-recurrence defined in equation (31) Simulation Results The implementation is illustrated for the double integrator and CSE examples, showing that all tube variants yield feasible solutions for specific MPC parameters The results show that the zonotopic variants outperform the polyhedral case, with Z+E(Φ) exhibiting a slight advantage over the more restrictive Z+E(Φ0), at the price of a higher computational burden The success of zonotopic implementations across various tube types and set representations confirms their efficacy for high-dimensional systems

Title and authors: Rosa: So to summarize what they did in "Zonotope-Based Elastic Tube Model Predictive Control," they took existing elastic tube MPC and introduced a novel linear way to encode zonotope containment.

Dev: They built on earlier work by extending the methods that were already doing elastic tube MPC, but the main goal was to avoid having to choose those auxiliary variables in advance.

Rosa: They showed this new encoding works for scaled zonotopes, and crucially, this approach eliminates the bilinear terms that usually pop up when you try to scale things in other methods.

Dev: That's important because it means the inclusion test itself remains linear with respect to both auxiliary variables and scaling factors, which makes it feasible for integration into large-scale optimization solvers.

Taro: From a researcher’s view, the paper seems focused on making the theoretical framework more practical by removing those preselection steps that usually slow down model building.

Rosa: Right. They also provided a way to do this in a computationally efficient precomputation-based alternative, which allows you to see exactly how much computational effort you're trading for how much bigger your domain of attraction gets.

Dev: That explicit quantification of the trade-off is what’s interesting for me; it gives us a concrete number to work with when we decide on our control frequency and complexity level.

Taro: It seems like they are pushing towards a system that can handle more complex uncertainty shapes while keeping the actual optimization problem tractable for real-time use.

Rosa: And they also established the conditions needed to guarantee recursive feasibility and asymptotic stability, which is the foundation for any robust control design in an MPC setting.

Dev: So they’re not just proposing a new way to state constraints; they are providing a complete set of tools to prove that the resulting system will actually stay stable and feasible over time.

Taro: It’s about bridging the gap between having a theoretically robust method and actually implementing it in a system where things might go sideways.

The paper's summary: Rosa: The authors suggest several improvements, starting with this novel linear encoding for zonotope containment which circumvents the need to preselect auxiliary variables.

Dev: That’s the core mechanism, and it extends naturally to scaled zonotopes without introducing those problematic bilinearities associated with scaling factors.

Taro: It seems like they’ve managed to generalize standard conditions in a way that keeps things linear, which is a major simplification for implementation.

Rosa: They also introduce a computationally efficient precomputation-based alternative, which lets you explicitly quantify the computational trade-offs you are making.

Dev: That precomputation idea is powerful because it allows the inclusion test to stay linear in both auxiliary variables and scaling factors, making it compatible with large optimization frameworks.

Taro: It means we can build these tubes up for complex scenarios without getting stuck on a combinatorial nightmare when defining the uncertainty sets.

Rosa: And they enhance the Robust Positively Invariant set approximation by treating those zonotope centers as independent decision variables, which is another practical step forward.

Dev: That move simplifies the problem structure by making those centers easier to handle within the optimization loop itself.

Taro: It suggests a path toward generating tube profiles that are more flexible and better suited for high-dimensional systems, rather than being stuck with overly restrictive polyhedral approximations.

The paper's improvements: Rosa: So, wrapping up the paper "Zonotope-Based Elastic Tube Model Predictive Control," the main point is that this new approach gives us a way to handle robust control tubes using scaled zonotopes more efficiently.

Dev: They managed to create a formulation where you can explicitly quantify the computational trade-off between how large your domain of attraction is and how much time you spend solving the optimization problem.

Taro: For me, it means we have a better tool for generating tube profiles that are more flexible for high-dimensional systems, which is what I need when testing real-world failure modes.

Rosa: And they provide solid guarantees on recursive feasibility and asymptotic stability through set recurrence definitions, which gives us confidence in the control system's long-term behavior.

Dev: They also show that this zonotopic approach outperforms some polyhedral methods in terms of computational time for certain models, even though it does require a higher effort than the simplest cases.

Taro: It’s a solid step forward because it moves us toward systems that can handle more complex uncertainty shapes without getting bogged down in overly restrictive approximations.

Rosa: That’s the gist of the paper. It shows that using scaled zonotopes for tube-based MPC can significantly reduce numerical complexity while still achieving robust performance.

Dev: Yeah, it sets a new direction for how we approach parameterizing uncertainty in these kinds of control problems, moving away from methods that require guessing constraints upfront.

Taro: We'll keep an eye on how this framework performs when we plug in the really messy, unstructured data from the real world.

Conclusion: Rosa: So we've been talking about "Zonotope-Based Elastic Tube Model Predictive Control," and basically, this paper shows how to use a new linear way to encode zonotope containment that lets us avoid having to pick those auxiliary variables in advance, which cuts down on the math a lot.

Dev: That’s the core of it, Rosa. They introduced this novel linear encoding for zonotope containment that generalizes standard conditions while avoiding those bilinear terms that usually pop up when you scale things up in other methods.

Taro: What I find interesting is how they made this fit into a larger framework, because the inclusion test itself remains linear with respect to both the auxiliary variables and the scaling factors, which makes it feasible for large optimization solvers.

Rosa: Exactly. And they show you how to do this in a computationally efficient way through precomputation that lets you see exactly what kind of computational trade-off you’re making when you want a bigger domain of attraction.

Dev: I appreciate that explicit quantification, Rosa, because it gives us a concrete number to work with when we decide on the control frequency and the complexity level for our real-time loops.

Taro: And from where I sit testing failure modes, they’ve established conditions to guarantee recursive feasibility and asymptotic stability by looking at set recurrence definitions which means we have some confidence in how the system will behave over time.

Rosa: So we’re talking about a framework that can handle high-dimensional systems with uncertainty shapes using scaled zonotopes while keeping the optimization problem tractable for real-time use.

Dev: It really does, and it moves us past methods that require you to preselect those constraints, which is where a lot of the complexity usually hides.

Taro: I think this is important because it gives us a better way to build control systems that can gracefully recover when the world misbehaves outside of the ideal test environment.

Rosa: That’s right. The paper "Zonotope-Based Elastic Tube Model Predictive Control" shows us how to make these robust frameworks more practical for real-world applications with high uncertainty.

Dev: It’s a solid piece of work that gives us better tools for designing control tubes under disturbance, and I think we should keep an eye on these kinds of scalable encodings.

Taro: Next up, we're going to look at how this affects online reinforcement learning when the environment is totally unpredictable.

Sabin Diaconescu, Florin Stoican, Bogdan D. Ciubotaru, Sorin Olaru

Department of Automatic Control and Systems Engineering, Nat. Univ. of Science and Technology Politehnica Bucharest · Laboratoire des Signaux et Systèmes, CentraleSup´elec, Paris-Saclay University

eess.SY, cs.SY

Submitted: 2025-09-24

Updated: 2026-10-30

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

Importance score: 73/100

The gist: Tube-based Model Predictive Control (MPC) is a widely adopted robust control framework for constrained linear systems under additive disturbance The paper focuses on reducing the numerical complexity…

Key concepts

Elastic Tube MPC
This is a control method designed for systems subject to disturbances. It works by defining an 'elastic tube' around the nominal system trajectory. The goal is to keep the actual system state within this tube, providing robustness against uncertainties while optimizing performance.
Scaled Zonotopes
Zonotopes are geometric shapes used to represent sets of possible states or disturbances. Scaling them allows for a more flexible representation of these uncertainty sets. This method uses scaled zonotopes to define the tube boundaries, which helps in generating tube profiles that lead to an enlarged domain of attraction.
Linear Encoding for Zonotope Containment
The paper introduces a new way to mathematically describe whether one set (like a zonotope) is contained within another. This linear encoding avoids the need to manually specify auxiliary variables beforehand, significantly reducing the complexity and effort required when setting up the control problem.
Computational Trade-offs Quantification
The formulation reformulates the optimization problem so that it explicitly shows how computational effort relates to performance gains. By making certain inclusion tests linear in both auxiliary variables and scaling factors, researchers can clearly see the balance between solving a complex optimization problem and achieving better system control.

Terminology

Summary

Tube-based Model Predictive Control (MPC) is a widely adopted robust control framework for constrained linear systems under additive disturbance The paper focuses on reducing the numerical complexity associated with tube parameterization, described as a sequence of elastically-scaled zonotopic sets A new class of scaled-zonotope inclusion conditions is proposed, alleviating the need for a priori specification of certain set-containment constraints and achieving significant reductions in complexity

Zonotope Fundamentals

Zonotopes are centrally symmetric polytopes that can be expressed as the Minkowski sum of line segments In their generator representation, a zonotope is defined as Z = ⟨c, G⟩ = ⟨x ∈ R n: x = c + Gξ,∥ξ∥∞ ≤ 1⟩ A natural generalization is the concept of a scaled zonotope defined by expressing G ← G∆, where ∆ = diag(δ), with δ ∈ R D ≥ 0

Scaled Zonotopic Inclusion Conditions

The paper introduces a novel linear encoding for zonotope containment that generalizes standard conditions, thereby circumventing the need to preselect auxiliary variables This formulation extends naturally to scaled zonotopes, eliminating the bilinearities typically associated with scaling factors and avoiding suboptimal parameter selection The proposed inclusion test remains linear in both auxiliary variables and scaling factors, facilitating its integration into large-scale optimization frameworks Corollary 1 shows that a sufficient condition to verify the inclusion Z1(δ1) ⊕... ⊕ Zn−1(δn−1) ⊆ Zn(δn) is given by a coefficient matrix Φ in R 2D×Dn This is linear in both the variable Φ and the scaling factors δi, i ∈ In

RPI Set Computation

The inclusion conditions (8)–(10) rely on the seedgenerator matrix G, which is commonly derived from the Robust Positive Invariant (RPI) set associated with the error dynamics A linear program is presented to compute a one-step zonotopic minimal RPI (mRPI) approximation, which involves minimizing c, Φ, δ to find the center c and scaling factor δ such that the zonotopic set c, G = G∆ is RPI The linear program of Prop. 3 involves 2sDω(1 + Dω + sDω) decision variables, fewer than the 3sDω(1 + Dω + sDω) variables used in [25, Prop. p. 4

Where the paper says it

✷ Remark 4 Denoting by Dω the number of generators in Gω, the linear program of Prop. 3 involves 2sDω(1 + Dω + sDω) decision variables, fewer than the 3sDω(1 + Dω + sDω) variables used in [25, Prop.

4]

Optimization Formulation and Complexity

The elastic-tube MPC problem is reformulated into a constrained quadratic-programming (QP) optimization problem where the goal is to find u¯⋆N, ΦN, δN, cN, λN This formulation verifies admissibility conditions (4)–(6) and recursive feasibility guarantees (26)–(28) Table 1 summarizes various scenarios for this problem, indicating the number of decision variables and constraints for different tube types and set types The zonotopic variants outperform the polyhedral case in terms of domain of attraction, with Z+E(Φ) exhibiting a slight advantage over the more restrictive Z+E(Φ0)

Simulation Results

The implementation is illustrated for the double integrator from [12] and the CSE example from [34] The results show that zonotopic variants outperform the polyhedral case, with Z+E(Φ) exhibiting a slight advantage over the more restrictive Z+E(Φ0), at the price of a higher computational burden For all cases, the inclusion of a disturbance-affected trajectory within its corresponding tube is verified The zonotopic implementations remain successful across the entire range of parameters, up to 20 states

Conclusions

The integration of elastically-scaled zonotopes into elastic-tube MPC significantly reduces runtime and improves scalability for high-dimensional systems The approach effectively manages additive disturbances in complex scenarios Future work will address the sensitivity of tube-based methods to the initial robust positively invariant set by reducing its complexity This model-based framework can be extended to data-driven MPC.

References

[1] J. Rawlings, D. p. 8

Where the paper says it

References [1] J. Rawlings, D.

Mayne, and M. p. 8
Where the paper says it

Rawlings, D. Mayne, and M.

Diehl, Model Predictive Control: Theory, Computation, and Design

[2] D. Q. Mayne, “Model predictive control: Recent developments and future promise,” Automatica

[3] F. Borrelli, A. p. 8

Where the paper says it

2967–2986, 2014. Borrelli, A.

Bemporad, and M. p. 8
Where the paper says it

Borrelli, A. Bemporad, and M.

Morari, Predictive Control for Linear and Hybrid Systems

[4] M. Morari and J. p. 8

Where the paper says it

Bemporad, and M. Morari, Predictive Control for Linear and Hybrid Systems.

H. Lee, “Model predictive control: past, present and future,” Computers & Chemical Engineering

[5] B. Kouvaritakis and M. p. 9

Where the paper says it

667–682, 1999. Kouvaritakis and M.

Cannon, Model Predictive Control: Classical, Robust and Stochastic

[6] F. Blanchini and S. p. 9

Where the paper says it

Springer International Publishing, 2016. Blanchini and S.

Miani, Set-theoretic methods in control

[7] D. Mayne, M. Seron, and S. p. 9

Where the paper says it

ISBN: 978-3-319-17933-9. Seron, and S.

Rakovi´c, “Robust model predictive control of constrained linear systems with bounded disturbances,” Automatica

[8] S. V. Rakovi´c, “The implicit rigid tube model predictive control,” Automatica

[9] S. V. Rakovi´c, B. p. 9

Where the paper says it

Seron, and S. Rakovi´c, “Robust model predictive control of constrained linear systems with bounded disturbances,” Automatica, vol.

Kouvaritakis, R. p. 9
Where the paper says it

667–682, 1999. Kouvaritakis and M.

Findeisen, and M. p. 9
Where the paper says it

Kouvaritakis, R. Findeisen, and M.

Cannon, “Homothetic tube model predictive control,” Automatica

[10] S. V. Rakovic, B. p. 9

Where the paper says it

1631–1638, 2012. Rakovic, B.

Kouvaritakis, M. p. 9
Where the paper says it

667–682, 1999. Kouvaritakis and M.

Cannon, C. Panos, and R. p. 9
Where the paper says it

Kouvaritakis, M. Panos, and R.

Findeisen, “Parameterized tube model predictive control,” IEEE Transactions on Automatic Control

[11] J. Fleming, B. p. 9

Where the paper says it

2746–2761, 2012. Fleming, B.

Kouvaritakis, and M. p. 9
Where the paper says it

667–682, 1999. Kouvaritakis and M.

Cannon, “Robust Tube MPC for Linear Systems With Multiplicative Uncertainty,” IEEE Transactions on Automatic Control

[12] S. V. Rakovi´c, W. p. 9

Where the paper says it

Seron, and S. Rakovi´c, “Robust model predictive control of constrained linear systems with bounded disturbances,” Automatica, vol.

S. Levine, and B. p. 9
Where the paper says it

Rakovi´c, W. Levine, and B.

A¸cikmese, “Elastic tube model predictive control,” in American Control Conference

[13] A. Parsi, A. Iannelli, and R. p. 9

Where the paper says it

3594–3599, 2016. Iannelli, and R.

S. Smith, “Scalable tube model predictive control of uncertain linear systems using ellipsoidal sets,” arXiv preprint arXiv:2204. p. 9
Where the paper says it

Iannelli, and R. Smith, “Scalable tube model predictive control of uncertain linear systems using ellipsoidal sets,” arXiv preprint arXiv:2204.

02134

[14] M. E. Villanueva, M. p. 9

Where the paper says it

02134, 2022. Villanueva, M.

A. M¨uller, and B. p. 9
Where the paper says it

Villanueva, M. M¨uller, and B.

Houska, “Configuration-constrained tube mpc,” Automatica

[15] B. S. Rego, G. V. Raffo, J. K. Scott, and D. p. 9

Where the paper says it

111543, 2024. Scott, and D.

M. Raimondo, “Guaranteed methods based on constrained zonotopes for set-valued state estimation of nonlinear discrete-time systems,” Automatica

[16] D. M. Raimondo, G. p. 9

Where the paper says it

Scott, and D. Raimondo, “Guaranteed methods based on constrained zonotopes for set-valued state estimation of nonlinear discrete-time systems,” Automatica, vol.

R. Marseglia, R. p. 9
Where the paper says it

Raimondo, G. Marseglia, R.

D. Braatz, and J. p. 9
Where the paper says it

Marseglia, R. Braatz, and J.

K. Scott, “Closed-loop input design for guaranteed fault diagnosis using set-valued observers,” Automatica

[17] R. Andrade, J. p. 9

Where the paper says it

107–117, 2016. Andrade, J.

E. Normey-Rico, and G. p. 9
Where the paper says it

Andrade, J. Normey-Rico, and G.

V. Raffo, “Tubebased model predictive control based on constrained zonotopes,” IEEE Access

[18] V. Raghuraman and J. p. 9

Where the paper says it

50100–50113, 2024. Raghuraman and J.

p. 9
Where the paper says it

50100–50113, 2024. Raghuraman and J.

p. 9
Where the paper says it

50100–50113, 2024. Raghuraman and J.

P. Koeln, “Tube-based robust MPC with adjustable uncertainty sets using zonotopes,” in 2021 American Control Conference (ACC)

[19] V. Raghuraman and J. P. Koeln, “Hierarchical MPC for coupled subsystems using adjustable tubes,” Automatica

[20] W. Wang and J. p. 9

Where the paper says it

110435, 2022. Wang and J.

P. Koeln, “Tube-based robust mpc for two-timescale systems using reduced-order models,” IEEE Control Systems Letters

[21] D. Ioan, S. Olaru, S.-I. Niculescu, I. p. 9

Where the paper says it

799–804, 2023. Niculescu, I.

Prodan, and F. p. 9
Where the paper says it

Niculescu, I. Prodan, and F.

Stoican, “Navigation in a multi-obstacle environment. p. 9
Where the paper says it

Prodan, and F. Stoican, “Navigation in a multi-obstacle environment.

from partition of the space to a zonotopic-based MPC,” in 18th European Control Conference

[22] L. Yang and N. p. 9

Where the paper says it

1772–1777, IEEE, 2019. Yang and N.

Ozay, “Scalable zonotopic underapproximation of backward reachable sets for uncertain linear systems,” IEEE Control Systems Letters

[23] V. Raghuraman and J. P. Koeln, “Set operations and order reductions for constrained zonotopes,” Automatica

[24] A. Richards and J. p. 9

Where the paper says it

110204, 2022. Richards and J.

How, “Robust stable model predictive control with constraint tightening,” in American Control Conference

[25] S. Diaconescu, F. p. 0

Where the paper says it

Zonotope-Based Elastic TubeModel Predictive Control ⋆ Sabin Diaconescu a , Florin Stoican a , Bogdan D. Ciubotaru a , Sorin Olaru b aDepartment of Automatic Control and Systems Engineering, Nat.

Stoican, B. p. 0
Where the paper says it

Zonotope-Based Elastic TubeModel Predictive Control ⋆ Sabin Diaconescu a , Florin Stoican a , Bogdan D. Ciubotaru a , Sorin Olaru b aDepartment of Automatic Control and Systems Engineering, Nat.

D. Ciubotaru, and S. p. 0
Where the paper says it

Zonotope-Based Elastic TubeModel Predictive Control ⋆ Sabin Diaconescu a , Florin Stoican a , Bogdan D. Ciubotaru a , Sorin Olaru b aDepartment of Automatic Control and Systems Engineering, Nat.

Olaru, “Elastic tube model predictive control with scaled zonotopic sets,” IEEE Control Systems Letters

[26] B. Houska, M. A. M¨uller, and M. p. 9

Where the paper says it

Villanueva, M. M¨uller, and B.

E. Villanueva, “Polyhedral control design: Theory and methods,” Annual Reviews in Control

[27] K. Fukuda, “Polyhedral computation,” Department of Mathematics, Institute of Theoretical Computer Science ETH Zurich

[28] B. Houska and M. p. 9

Where the paper says it

M¨uller, and B. Houska, “Configuration-constrained tube mpc,” Automatica, vol.

E. Villanueva, “Robust Optimization for MPC,” in Handbook of Model Predictive Control (S. p. 9
Where the paper says it

Houska and M. Villanueva, “Robust Optimization for MPC,” in Handbook of Model Predictive Control (S.

V. Rakovi´c and W. p. 9
Where the paper says it

Seron, and S. Rakovi´c, “Robust model predictive control of constrained linear systems with bounded disturbances,” Automatica, vol.

S. Levine, eds. p. 9
Where the paper says it

Rakovi´c and W. Levine, eds.

), Springer International Publishing

[29] S. Sadraddini and R. p. 9

Where the paper says it

413–443, Springer International Publishing, 2019. Sadraddini and R.

Tedrake, “Linear encodings for polytope containment problems,” in 2019 IEEE 58th Conference on Decision and Control (CDC)

[30] S. V. Rakovic, E. p. 9

Where the paper says it

1631–1638, 2012. Rakovic, B.

C. Kerrigan, K. p. 9
Where the paper says it

Rakovic, E. Kerrigan, K.

I. Kouramas, and D. p. 9
Where the paper says it

Kerrigan, K. Kouramas, and D.

Q. Mayne, “Invariant approximations of the minimal robust positively invariant set,” IEEE Transactions on automatic control

[31] D. Mayne, J. Rawlings, C. p. 8

Where the paper says it

References [1] J. Rawlings, D.

Rao, and P. p. 9
Where the paper says it

Rawlings, C. Rao, and P.

Scokaert, “Constrained model predictive control: Stability and optimality,” Automatica

[32] H. Chen and F. p. 9

Where the paper says it

789–814, 2000. Chen and F.

Allg¨ower, “A quasi-infinite horizon nonlinear model predictive control scheme with guaranteed stability,” Automatica

[33] S. V. Rakovi´c and S. p. 9

Where the paper says it

Seron, and S. Rakovi´c, “Robust model predictive control of constrained linear systems with bounded disturbances,” Automatica, vol.

Zhang, “The implicit maximal positively invariant set,” IEEE Transactions on Automatic Control

[34] F. Leibfritz, “COMPleib: COnstrained Matrix–optimization Problem library – a collection of test examples for nonlinear semidefinite programs, control system design and related problems,” 2006

[35] J. L¨ofberg, “Automatic robust convex programming,” Optimization methods and software

[36] M. Althoff, “An Introduction to CORA 2015,” in Proc. p. 9

Where the paper says it

115– 129, 2012. Althoff, “An Introduction to CORA 2015,” in Proc.

of the 1st and 2nd Workshop on Applied Verification for Continuous and Hybrid Systems, pp. p. 9
Where the paper says it

Althoff, “An Introduction to CORA 2015,” in Proc. of the 1st and 2nd Workshop on Applied Verification for Continuous and Hybrid Systems, pp.

120–151, December 2015

[37] A. Chalkis, V. p. 9

Where the paper says it

120–151, December 2015. Chalkis, V.

Fisikopoulos, M. p. 9
Where the paper says it

Chalkis, V. Fisikopoulos, M.

Papachristou, and E. p. 9
Where the paper says it

Fisikopoulos, M. Papachristou, and E.

Tsigaridas, “volesti: A C++ library for sampling and volume computation on convex bodies,” Journal of Open Source Software

[38] J. Hespanha, “Tenscalc: a toolbox to generate fast code to solve nonlinear constrained minimizations and compute Nash equilibria,” Mathematical Programming Computation

[39] B. Stellato, G. p. 9

Where the paper says it

451–496, 2022. Stellato, G.

Banjac, P. Goulart, A. p. 9
Where the paper says it

Stellato, G. Goulart, A.

Bemporad, and S. p. 8
Where the paper says it

Borrelli, A. Bemporad, and M.

Boyd, “OSQP: an operator splitting solver for quadratic programs,” Mathematical Programming Computation

[40] M. Farjadnia, A. p. 9

Where the paper says it

637–672, 2020. Farjadnia, A.

Fontan, A. Alanwar, M. p. 9
Where the paper says it

Farjadnia, A. Alanwar, M.

Molinari, and K. p. 9
Where the paper says it

Alanwar, M. Molinari, and K.

H.

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

  1. Robust Model Predictive Control for High-Dimensional Systems: The improved system can perform robust control of linear time-invariant systems with additive disturbances by employing elastic tube model predictive control, which is specifically designed to reduce numerical complexity associated with set parameterization.

  2. Reduced Complexity in Tube Parameterization: The system can utilize a novel linear encoding for zonotope containment that circumvents the need to preselect auxiliary variables, thereby alleviating the need for a priori specification of certain set-containment constraints and achieving significant reductions in complexity.

  3. Scalable Zonotopic Tube Profile Generation: The improved system can generate tube profiles using scaled zonotopes where constant scaling factors throughout the prediction horizon are enforced, allowing for an enlarged domain of attraction and improved performance.

  4. Computationally Efficient Precomputation: The system can utilize a precomputation-based alternative that allows for an explicit quantification of computational trade-offs by reformulating the problem such that the proposed inclusion test remains linear in both auxiliary variables and scaling factors, facilitating its integration into large-scale optimization frameworks.

  5. State and Input Admissibility Verification: The system can rigorously verify state and input admissibility conditions (5)–(6) by ensuring constraints like FX(ck + ¯xk) + FXG δk ≤ θX are respected for all subsequent iterations of the closed-loop extended dynamics.

  6. Recursive Feasibility and Stability Guarantees: The improved system can guarantee recursive feasibility and asymptotic stability by verifying Lyapunov-like contraction conditions, such as VT (¯xN+1, δN+1) − VT (¯xN, δN) + l(¯xN, KT X¯N, δN) ≤ 0.

  7. Efficient Optimization Formulation: The system can solve the elastic-tube MPC problem by reformulating it into a constrained quadratic programming problem where the decision variables include nominal states, scaling factors, and auxiliary variables like c and lambda.

Abstract

Tube-based Model Predictive Control (MPC) is a widely adopted robust control framework for constrained linear systems under additive disturbance. The paper is focused on reducing the numerical complexity associated with the tube parameterization, described as a sequence of elastically-scaled zonotopic sets. A new class of scaled-zonotope inclusion conditions is proposed, alleviating the need for a priori specification of certain set-containment constraints and achieving significant reductions in complexity. A comprehensive complexity analysis is provided for both the polyhedral and the zonotopic setting, illustrating the trade-off between an enlarged domain of attraction and the required computational effort. The proposed approach is validated through extensive numerical experiments.

Sources

Related papers