pith. sign in

arxiv: 2604.25766 · v2 · submitted 2026-04-28 · 💻 cs.RO

Sensitivity-Based Tube NMPC for Cooperative Aerial Structures Under Parametric Uncertainty

Giuseppe Silano , Quentin Sabl\'e , Marco Tognon , Luigi Iannelli , Antonio Franchi This is my paper

Pith reviewed 2026-05-07 15:40 UTC · model grok-4.3

classification 💻 cs.RO
keywords tube NMPCparametric uncertaintysensitivity propagationaerial chainscooperative roboticsconstraint tighteningrobust controlrigid links
0
0 comments X

The pith

Propagating first-order state sensitivities with respect to link parameters produces online tightening margins that keep separation and thrust constraints satisfied in aerial chains despite bounded mass, length, and inertia uncertainty.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper develops a tube NMPC method for a planar two-vehicle aerial chain connected by rigid links. It propagates first-order sensitivities of the predicted states to uncertain parameters along the horizon to compute time-varying constraint margins in real time. These margins robustify a smooth cosine embedding of the inter-link separation distance and the thrust-magnitude bounds. The resulting controller is tested on boundary-hugging trajectories with Monte-Carlo sampling of the uncertain parameters. Tracking accuracy stays comparable to nominal NMPC while constraint satisfaction improves under the modeled uncertainty.

Core claim

The central claim is that first-order parametric state sensitivities, propagated along the NMPC horizon, can be used to compute online constraint-tightening margins that guarantee satisfaction of the inter-vehicle separation constraint and actuator bounds for any realization of the bounded parametric uncertainty in link mass, length, and inertia.

What carries the argument

First-order propagation of parametric state sensitivities used to derive time-varying tightening margins in a tube NMPC formulation.

If this is right

  • The inter-link separation constraint, embedded via a smooth cosine function, and the thrust-magnitude bounds become robust to the modeled parametric uncertainty.
  • Constraint margins are computed online rather than fixed offline, allowing the controller to adapt tightening to the current state and reference.
  • Tracking performance remains comparable to a nominal NMPC while constraint satisfaction improves under uncertainty.
  • Input-rate actuation is used throughout to enforce both magnitude and slew-rate limits on thrust and torque.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same sensitivity-propagation mechanism could be applied to chains of three or more vehicles provided the first-order model stays valid.
  • The approach might reduce conservatism relative to methods that compute a single fixed tube radius for the entire uncertainty set.
  • Because margins are updated at each step, the method could be combined with adaptive parameter estimation to further shrink the uncertainty bounds online.

Load-bearing premise

The first-order approximation of state sensitivities is accurate enough across the prediction horizon to keep the tightened constraints from being violated by the actual bounded parametric uncertainty.

What would settle it

A Monte-Carlo trial in which the closed-loop system with parameter values drawn from the uncertainty bounds produces either a separation-distance violation or a thrust-magnitude violation during a boundary-hugging maneuver.

Figures

Figures reproduced from arXiv: 2604.25766 by Antonio Franchi, Giuseppe Silano, Luigi Iannelli, Marco Tognon, Quentin Sabl\'e.

Figure 7
Figure 7. Figure 7: – Representation of the system and its main variables. The systemRepresentation of thesystem and its main variables. Thesystem Fig. 1: Representation of the cooperative aerial structure and iiiblThidid il view at source ↗
Figure 2
Figure 2. Figure 2: Boundary-hugging maneuver under parametric uncer view at source ↗
Figure 6
Figure 6. Figure 6: Signed constraint residuals for nominal NMPC (top) view at source ↗
Figure 7
Figure 7. Figure 7: Boxplots for the Monte-Carlo campaign, comparing view at source ↗
Figure 5
Figure 5. Figure 5: Tracking errors for the trajectories in Figure 4: view at source ↗
read the original abstract

This paper presents a sensitivity-based tube Nonlinear Model Predictive Control (NMPC) framework for cooperative aerial chains under bounded parametric uncertainty. We consider a planar two-vehicle chain connected by rigid links, modeled with input-rate actuation to enforce slew-rate and magnitude limits on thrust and torque. Robustness to uncertainty in link mass, length, and inertia is achieved by propagating first-order parametric state sensitivities along the horizon and using them to compute online constraint-tightening margins. We robustify an inter-link separation constraint, implemented via a smooth cosine embedding, and thrust-magnitude bounds. The method is implemented in MATLAB and evaluated with boundary-hugging maneuvers and Monte-Carlo uncertainty sampling. Results show improved constraint margins under uncertainty with tracking performance comparable to nominal NMPC.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 2 minor

Summary. The paper presents a sensitivity-based tube NMPC framework for a planar two-vehicle aerial chain under bounded parametric uncertainty in link mass, length, and inertia. Robustness is obtained by propagating first-order parametric state sensitivities along the horizon to compute online constraint-tightening margins for an inter-link cosine-embedded separation constraint and thrust-magnitude bounds. The system model incorporates input-rate actuation. The approach is implemented in MATLAB and evaluated on boundary-hugging maneuvers using Monte-Carlo uncertainty sampling, with results indicating improved constraint margins relative to nominal NMPC while preserving comparable tracking performance.

Significance. If the first-order sensitivity approximation proves sufficiently accurate, the method supplies a practical, online-computable mechanism for robustifying NMPC on uncertain cooperative aerial structures without requiring full reachable-set propagation. The Monte-Carlo validation and MATLAB implementation provide empirical support for the practical utility of the tightening margins, representing a modest but concrete contribution to real-time robust control of multi-agent aerial systems.

major comments (1)
  1. The central robustness claim rests on propagating first-order parametric state sensitivities to obtain constraint-tightening margins (abstract and sensitivity-based tube NMPC description). No remainder bound, Lipschitz constant on the sensitivity map, or re-linearization procedure is supplied to control the error when parametric uncertainty drives the actual state away from the nominal linearization trajectory. In the nonlinear rigid-link dynamics this omission is load-bearing, as the computed margins may fail to enclose the true reachable set and therefore do not rigorously guarantee satisfaction of the original inter-link and thrust constraints.
minor comments (2)
  1. The abstract states that results show 'improved constraint margins' but supplies no quantitative values, uncertainty ranges, or statistical summaries from the Monte-Carlo trials; adding these numbers would strengthen the empirical claim.
  2. The choice of a smooth cosine embedding for the inter-link separation constraint is mentioned without discussion of alternatives (e.g., direct distance or barrier functions) or the effect of the embedding parameter on the sensitivity computation.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive review, positive assessment of the method's practical utility, and recommendation for major revision. We address the single major comment below.

read point-by-point responses

  1. Referee: The central robustness claim rests on propagating first-order parametric state sensitivities to obtain constraint-tightening margins (abstract and sensitivity-based tube NMPC description). No remainder bound, Lipschitz constant on the sensitivity map, or re-linearization procedure is supplied to control the error when parametric uncertainty drives the actual state away from the nominal linearization trajectory. In the nonlinear rigid-link dynamics this omission is load-bearing, as the computed margins may fail to enclose the true reachable set and therefore do not rigorously guarantee satisfaction of the original inter-link and thrust constraints.

    Authors: We agree that the manuscript relies on a first-order sensitivity approximation for online margin computation without supplying an explicit remainder bound, Lipschitz constant, or re-linearization scheme. The paper does not claim a rigorous enclosure of the reachable set under the nonlinear rigid-link dynamics; instead, it presents the approach as a computationally lightweight heuristic whose practical robustness is supported by Monte-Carlo validation on boundary-hugging maneuvers. We will revise the manuscript to (i) explicitly state in the abstract and method sections that the tightening margins are first-order approximations without formal error guarantees, (ii) add a dedicated paragraph discussing the approximation error sources and the conditions under which the method remains effective, and (iii) outline possible future extensions such as periodic re-linearization or conservative Lipschitz-based over-approximations. This constitutes a partial revision that clarifies scope while preserving the empirical contribution. revision: partial

Circularity Check

0 steps flagged

No circularity: sensitivity propagation is an independent algorithmic step, not a self-referential fit or definition.

full rationale

The paper's core construction propagates first-order parametric state sensitivities along a nominal trajectory to obtain online tightening margins for cosine-embedded inter-link and thrust constraints. This step is defined directly from the system dynamics and uncertainty bounds without reducing to a fitted parameter renamed as a prediction, a self-definitional loop, or a load-bearing self-citation whose validity depends on the present result. The method is presented as a direct application of standard sensitivity analysis inside tube NMPC; the derivation chain remains self-contained against external benchmarks and does not import uniqueness or ansatz via prior author work in a circular manner.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The approach builds on standard NMPC and sensitivity propagation techniques without introducing new free parameters or entities in the abstract description.

axioms (2)
  • domain assumption The system dynamics are differentiable with respect to parameters
    Required for propagating first-order sensitivities.
  • domain assumption Uncertainty is bounded
    For tube MPC robustness.

pith-pipeline@v0.9.0 · 5431 in / 1132 out tokens · 52081 ms · 2026-05-07T15:40:19.187970+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

21 extracted references · 21 canonical work pages

  1. [1]

    Past, Present, and Future of Aerial Robotic Manipulators,

    A. Ollero, M. Tognon, A. Suarez, D. Lee, and A. Franchi, “Past, Present, and Future of Aerial Robotic Manipulators,”IEEE Transac- tions on Robotics, vol. 38, no. 1, pp. 626–645, 2022

    work page 2022

  2. [2]

    Aerial Manipulation: A Literature Review,

    F. Ruggiero, V . Lippiello, and A. Ollero, “Aerial Manipulation: A Literature Review,”IEEE Robotics and Automation Letters, vol. 3, no. 3, pp. 1957–1964, 2018

    work page 1957

  3. [3]

    Multi-Aerial Robotic System for Power Line Inspection and Maintenance: Com- parative Analysis From the AERIAL-CORE Final Experiments,

    A. Ollero, A. Suarez, C. Papaioannidis, I. Pitas, J. M. Marredo, V . D. Hoang, E. Ebeid, V . Kratky, M. Saska, C. Hanoune, A. Afifi, A. Franchi, C. V ourtsis, D. Floreano, G. Vasiljevic, S. Bogdan, A. Caballero, F. Ruggiero, V . Lippiello, C. Matilla, G. Cioffi, D. Scara- muzza, J. R. Martinez de Dios, B. C. Arrue, C. Martin, K. Zurad, C. Gaitan, J. Rodri...

    work page 2025

  4. [4]

    Multi-directional Interaction Force Control with an Aerial Manipu- lator Under External Disturbances,

    G. Malczyk, M. Brunner, E. Cuniato, M. Tognon, and R. Siegwart, “Multi-directional Interaction Force Control with an Aerial Manipu- lator Under External Disturbances,”Autonomous Robots, vol. 47, pp. 1325–1343, 2023

    work page 2023

  5. [5]

    Cooperative Aerial Ma- nipulation Using Multirotors With Multi-DOF Robotic Arms,

    S. Kim, H. Seo, J. Shin, and H. J. Kim, “Cooperative Aerial Ma- nipulation Using Multirotors With Multi-DOF Robotic Arms,”IEEE Transactions on Mechatronics, vol. 23, no. 2, pp. 702–713, 2018

    work page 2018

  6. [6]

    Geometric Control of Two Quadrotors Carrying a Rigid Rod with Elastic Cables,

    J. Goodman and L. Colombo, “Geometric Control of Two Quadrotors Carrying a Rigid Rod with Elastic Cables,”Journal of Nonlinear Science, vol. 32, no. 65, 2022. 4https://docs.acados.org/

    work page 2022

  7. [7]

    Control of motion and internal stresses for a chain of two underactuated aerial robots,

    M. Tognon and A. Franchi, “Control of motion and internal stresses for a chain of two underactuated aerial robots,” inEuropean Control Conference, 2015, pp. 1620–1625

    work page 2015

  8. [8]

    LASDRA: Large-Size Aerial Skeleton System with Distributed Rotor Actuation,

    H. Yang, S. Park, J. Lee, J. Ahn, D. Son, and D. Lee, “LASDRA: Large-Size Aerial Skeleton System with Distributed Rotor Actuation,” inIEEE International Conference on Robotics and Automation, 2018, pp. 7017–7023

    work page 2018

  9. [9]

    Gr ¨une and J

    L. Gr ¨une and J. Pannek,Nonlinear Model Predictive Control: Theory and Algorithms, 2nd ed. Springer, 2017

    work page 2017

  10. [10]

    J. B. Rawlings, D. Q. Mayne, and M. M. Diehl,Model Predictive Control: Theory, Computation, and Design, 2nd ed. Nob Hill Publishing, 2022

    work page 2022

  11. [11]

    A robust adaptive model predictive control framework for nonlinear uncertain systems,

    J. Kohler, P. Kotting, R. Soloperto, F. Allgower, and M. A. Muller, “A robust adaptive model predictive control framework for nonlinear uncertain systems,”International Journal of Robust and Nonlinear Control, vol. 31, no. 18, pp. 8725–8749, 2020

    work page 2020

  12. [12]

    Robust model predictive control using tubes,

    W. Langson, I. Chryssochoos, S. V . Rakovic, and D. Q. Mayne, “Robust model predictive control using tubes,”Automatica, vol. 40, no. 1, pp. 125–133, 2004

    work page 2004

  13. [13]

    On the design of Robust tube-based MPC for tracking,

    D. Limon, I. Alvarado, T. Alamo, and E. Camacho, “On the design of Robust tube-based MPC for tracking,” inIF AC World Congress, vol. 41, no. 2, 2008, pp. 15 333–15 338

    work page 2008

  14. [14]

    Embedded Robust Model Predictive Path Integral Control Using Sensitivity Tubes and GPU Acceleration,

    F. F. Nyboe, A. Afifi, P. Robuffo Giordano, E. Ebeid, and A. Franchi, “Embedded Robust Model Predictive Path Integral Control Using Sensitivity Tubes and GPU Acceleration,” inIEEE International Conference on Robotics and Automation, 2025, pp. 7174–7180

    work page 2025

  15. [15]

    Real- Time Tube MPC Applied to a 10-State Quadrotor Model,

    H. Hu, X. Feng, R. Quirynen, M. E. Villanueva, and B. Houska, “Real- Time Tube MPC Applied to a 10-State Quadrotor Model,” inIEEE American Control Conference, 2018, pp. 3135–3140

    work page 2018

  16. [16]

    Trajectory Generation for Minimum Closed-Loop State Sensitivity,

    P. Robuffo Giordano, Q. Delamare, and A. Franchi, “Trajectory Generation for Minimum Closed-Loop State Sensitivity,” inIEEE International Conference on Robotics and Automation, 2018, pp. 286– 293

    work page 2018

  17. [17]

    Robust Model Predictive Control for Nonlinear Discrete-Time Systems using Iterative Time-Varying Constraint Tightening,

    D. D. Leister and J. P. Koeln, “Robust Model Predictive Control for Nonlinear Discrete-Time Systems using Iterative Time-Varying Constraint Tightening,” inIEEE American Control Conference, 2025, pp. 71–78

    work page 2025

  18. [18]

    Sensitivity-Aware Model Predictive Control for Robots With Para- metric Uncertainty,

    T. Belvedere, M. Cognetti, G. Oriolo, and P. Robuffo Giordano, “Sensitivity-Aware Model Predictive Control for Robots With Para- metric Uncertainty,”IEEE Transactions on Robotics, vol. 41, pp. 3039–3058, 2025

    work page 2025

  19. [19]

    MPC on manifolds with an application to the control of spacecraft attitude on SO(3),

    U. V . Kalabi ´c, R. Gupta, S. Di Cairano, A. M. Bloch, and I. V . Kolmanovsky, “MPC on manifolds with an application to the control of spacecraft attitude on SO(3),”Automatica, vol. 76, pp. 293–300, 2017

    work page 2017

  20. [20]

    Non- linear Model Predictive Control with Enhanced Actuator Model for Multi-Rotor Aerial Vehicles with Generic Designs,

    D. Bicego, J. Mazzetto, R. Carli, M. Farina, and A. Franchi, “Non- linear Model Predictive Control with Enhanced Actuator Model for Multi-Rotor Aerial Vehicles with Generic Designs,”Journal of Intel- ligent & Robotic Systems, vol. 100, pp. 1213–1247, 2020

    work page 2020

  21. [21]

    Nonlinear Model Predictive Control for Aerial Manipulation,

    D. Lunni, A. Santamaria-Navarro, R. Rossi, P. Rocco, L. Bascetta, and J. Andrade-Cetto, “Nonlinear Model Predictive Control for Aerial Manipulation,” inInternational Conference on Unmanned Aircraft Systems, 2017, pp. 87–93

    work page 2017