pith. machine review for the scientific record. sign in

arxiv: 2604.03200 · v1 · submitted 2026-04-03 · 💻 cs.RO · math.OC

Recognition: no theorem link

Safety-Critical Centralized Nonlinear MPC for Cooperative Payload Transportation by Two Quadrupedal Robots

Ruturaj S. Sambhus , Yicheng Zeng , Kapi Ketan Mehta , Jeeseop Kim , Kaveh Akbari Hamed
Authors on Pith no claims yet

Pith reviewed 2026-05-13 18:41 UTC · model grok-4.3

classification 💻 cs.RO math.OC
keywords nonlinear model predictive controlcontrol barrier functionsquadrupedal robotscooperative payload transportationsafety-critical controldifferential-algebraic equationscollision avoidancehardware experiments
0
0 comments X

The pith

A CBF-augmented centralized NMPC on a nonlinear DAE model enables two quadrupedal robots to transport a payload safely through cluttered spaces under uncertainties.

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

The paper develops a safety-critical centralized nonlinear model predictive control framework for cooperative payload transportation by two quadrupedal robots. The interconnected system is modeled as a discrete-time nonlinear differential-algebraic system that captures coupling through holonomic constraints and interaction wrenches. Control barrier functions are embedded directly into the NMPC to enforce collision avoidance for both the robots and the payload. Retaining the interaction wrenches as decision variables produces a structured optimization that supports real-time execution. Hardware validation on two Unitree Go2 robots shows successful operation in cluttered environments despite payload mass and inertia changes plus external disturbances.

Core claim

The central claim is that a CBF-based NMPC formulation on a discrete-time nonlinear DAE model of the coupled robot-payload dynamics, with interaction wrenches kept as decision variables, produces collision-free trajectories that remain safe and feasible under model uncertainty and disturbances.

What carries the argument

CBF-augmented NMPC on a discrete-time nonlinear DAE model that retains interaction wrenches as decision variables to enforce safety constraints efficiently.

If this is right

  • Collision avoidance constraints for robots and payload are satisfied in real time.
  • Stability is preserved when payload mass and inertia differ from the model.
  • External push disturbances are rejected without violating safety.
  • The formulation runs fast enough for direct deployment on quadruped hardware.
  • The approach works in cluttered indoor settings with static obstacles.

Where Pith is reading between the lines

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

  • The same DAE-plus-CBF structure could be extended to three or more robots for heavier or larger payloads.
  • Integration with online mapping would allow the method to handle previously unknown obstacle layouts.
  • Similar centralized safety-critical MPC might transfer to other multi-robot legged tasks such as collaborative object pushing.
  • Testing under dynamic obstacles or on uneven terrain would check whether the current constraint formulation remains sufficient.

Load-bearing premise

The robot-payload system dynamics are accurately represented by a discrete-time nonlinear differential-algebraic model with known holonomic constraints.

What would settle it

A hardware run in which a robot or the payload collides with an obstacle or the payload is lost while the NMPC is active under the stated mass uncertainty and push disturbances would falsify the safety claim.

Figures

Figures reproduced from arXiv: 2604.03200 by Jeeseop Kim, Kapi Ketan Mehta, Kaveh Akbari Hamed, Ruturaj S. Sambhus, Yicheng Zeng.

Figure 1
Figure 1. Figure 1: Snapshot of two Unitree Go2 quadrupedal robots cooperatively [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Overview of the proposed layered control framework. The high-level CBF-based NMPC computes optimal trajectories for the interconnected [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 4
Figure 4. Figure 4: Illustration of the rigid mechanism between the robots and [PITH_FULL_IMAGE:figures/full_fig_p003_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Snapshots of all experiments: (a) Experiment 1 with a nominal payload of 5 kg; (b) Experiment 2 with an unmodeled payload of 11.2 kg; (c) [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Reference and actual CoM velocity trajectories along the [PITH_FULL_IMAGE:figures/full_fig_p006_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: CoM trajectories of the robotic agents and the shared payload in [PITH_FULL_IMAGE:figures/full_fig_p007_7.png] view at source ↗
read the original abstract

This paper presents a safety-critical centralized nonlinear model predictive control (NMPC) framework for cooperative payload transportation by two quadrupedal robots. The interconnected robot-payload system is modeled as a discrete-time nonlinear differential-algebraic system, capturing the coupled dynamics through holonomic constraints and interaction wrenches. To ensure safety in complex environments, we develop a control barrier function (CBF)-based NMPC formulation that enforces collision avoidance constraints for both the robots and the payload. The proposed approach retains the interaction wrenches as decision variables, resulting in a structured DAE-constrained optimal control problem that enables efficient real-time implementation. The effectiveness of the algorithm is validated through extensive hardware experiments on two Unitree Go2 platforms performing cooperative payload transportation in cluttered environments under mass and inertia uncertainty and external push disturbances.

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

2 major / 2 minor

Summary. The paper presents a safety-critical centralized nonlinear model predictive control (NMPC) framework for cooperative payload transportation by two quadrupedal robots. The interconnected robot-payload system is modeled as a discrete-time nonlinear differential-algebraic system capturing coupled dynamics via holonomic constraints and interaction wrenches retained as decision variables. A control barrier function (CBF)-based NMPC formulation enforces collision avoidance for robots and payload. The approach is validated through hardware experiments on two Unitree Go2 platforms in cluttered environments under mass/inertia uncertainty and external push disturbances.

Significance. If the central claims hold, the work advances real-time safety-critical control for multi-robot cooperative manipulation tasks. Retaining wrenches in the DAE-constrained NMPC enables structured, efficient optimization suitable for onboard implementation. Hardware validation under realistic uncertainties and disturbances provides practical evidence of applicability to logistics or search-and-rescue scenarios, extending standard NMPC/CBF methods to interconnected DAE systems.

major comments (2)
  1. Modeling section (DAE formulation): Retaining interaction wrenches as decision variables incorporates holonomic and wrench-balance constraints into the decision space, but no analysis is provided of how mass/inertia mismatches (claimed 10-20% in experiments) propagate into algebraic constraint consistency or CBF gradient feasibility. Without robustness margins or tube-based tightening, model error can render the problem infeasible or violate safety guarantees even if the solver reports success.
  2. Experimental validation section: The hardware results claim safety enforcement under uncertainty and disturbances, but lack quantitative metrics (e.g., minimum CBF values, violation rates, or feasibility margins), baseline comparisons to standard NMPC, or error analysis to substantiate the real-time safety claim.
minor comments (2)
  1. Notation consistency: Ensure uniform symbols for interaction wrenches and constraint Jacobians across the DAE model and NMPC formulation to improve readability.
  2. Figure clarity: Hardware experiment figures would benefit from overlaid CBF values or constraint violation indicators to visually support the safety claims.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive and insightful comments on our manuscript. We address each major comment point by point below, indicating the revisions we plan to incorporate. These changes will strengthen the presentation of the DAE modeling and the experimental validation of safety guarantees.

read point-by-point responses

  1. Referee: Modeling section (DAE formulation): Retaining interaction wrenches as decision variables incorporates holonomic and wrench-balance constraints into the decision space, but no analysis is provided of how mass/inertia mismatches (claimed 10-20% in experiments) propagate into algebraic constraint consistency or CBF gradient feasibility. Without robustness margins or tube-based tightening, model error can render the problem infeasible or violate safety guarantees even if the solver reports success.

    Authors: We agree that a dedicated analysis of model mismatch effects on algebraic constraint consistency and CBF feasibility would improve the manuscript. The current formulation treats the DAE as exact for the nominal model, with interaction wrenches retained as decision variables to preserve structure and enable efficient solving. While hardware experiments under 10-20% uncertainty demonstrate practical performance, we acknowledge the lack of formal propagation analysis. In the revision, we will add a subsection in the modeling section (with supporting numerical simulations) that examines sensitivity of the holonomic constraints and CBF gradients to inertia/mass errors, including feasibility margin statistics. We note that the real-time re-optimization inherent to NMPC provides adaptability not captured by static tube methods, but we will discuss this explicitly. revision: yes

  2. Referee: Experimental validation section: The hardware results claim safety enforcement under uncertainty and disturbances, but lack quantitative metrics (e.g., minimum CBF values, violation rates, or feasibility margins), baseline comparisons to standard NMPC, or error analysis to substantiate the real-time safety claim.

    Authors: We concur that quantitative metrics and comparisons are necessary to rigorously substantiate the safety claims. The manuscript currently reports qualitative success in cluttered environments under disturbances but does not include aggregated statistics. In the revised manuscript, we will augment the experimental section with: (i) tables of minimum CBF values and violation rates across all trials, (ii) solver feasibility margins (e.g., constraint slack statistics), (iii) a direct baseline comparison against standard NMPC without CBF constraints, showing differences in collision avoidance rates, and (iv) error analysis on mass/inertia estimates and external disturbance magnitudes. These additions will be supported by additional plots and statistical summaries from the existing hardware data. revision: yes

Circularity Check

0 steps flagged

No circularity: standard DAE-NMPC-CBF formulation with independent validation

full rationale

The derivation models the robot-payload system as a discrete-time nonlinear DAE with holonomic constraints and retained interaction wrenches as decision variables, then applies a standard CBF-augmented NMPC formulation for collision avoidance. No step reduces by construction to fitted parameters, self-definitions, or self-citation chains; the central feasibility and safety claims rest on the explicit optimization problem and hardware experiments under uncertainty, which are externally falsifiable. The approach cites standard NMPC and CBF literature without load-bearing self-references that would force the result.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on standard robotics modeling assumptions for coupled dynamics and constraints; no new entities or heavily fitted parameters are introduced in the abstract.

axioms (1)
  • domain assumption The robot-payload system dynamics can be represented as a discrete-time nonlinear DAE with holonomic constraints.
    Invoked to model the interconnected system and enable the structured optimal control problem.

pith-pipeline@v0.9.0 · 5454 in / 1166 out tokens · 64399 ms · 2026-05-13T18:41:20.270040+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

35 extracted references · 35 canonical work pages

  1. [1]

    Templates and anchors: Neuromechanical hypotheses of legged locomotion on land,

    R. Full and D. Koditschek, “Templates and anchors: Neuromechanical hypotheses of legged locomotion on land,”Journal of Experimental Biology, vol. 202, no. 23, pp. 3325–3332, 1999

    work page 1999

  2. [2]

    Study of dynamic biped locomotion on rugged terrain-derivation and application of the linear inverted pendulum mode,

    S. Kajita and K. Tani, “Study of dynamic biped locomotion on rugged terrain-derivation and application of the linear inverted pendulum mode,” inIEEE International Conference on Robotics and Automation, 1991, pp. 1405–1406

    work page 1991

  3. [3]

    Terrain- adaptive, ALIP-based bipedal locomotion controller via model pre- dictive control and virtual constraints,

    G. Gibson, O. Dosunmu-Ogunbi, Y . Gong, and J. Grizzle, “Terrain- adaptive, ALIP-based bipedal locomotion controller via model pre- dictive control and virtual constraints,” inIEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2022, pp. 6724– 6731

    work page 2022

  4. [4]

    Compliant leg behavior explains basic dynamics of walking and running,

    H. Geyer, A. Seyfarth, and R. Blickhan, “Compliant leg behavior explains basic dynamics of walking and running,”Proceedings. Bi- ological sciences / The Royal Society, vol. 273, pp. 2861–7, 08 2006

    work page 2006

  5. [5]

    Autonomous navigation of underactuated bipedal robots in height-constrained environments,

    Z. Li, J. Zeng, S. Chen, and K. Sreenath, “Autonomous navigation of underactuated bipedal robots in height-constrained environments,” The International Journal of Robotics Research, vol. 42, no. 8, pp. 565–585, 2023

    work page 2023

  6. [6]

    3-D underactuated bipedal walking via H-LIP based gait synthesis and stepping stabilization,

    X. Xiong and A. Ames, “3-D underactuated bipedal walking via H-LIP based gait synthesis and stepping stabilization,”IEEE Transactions on Robotics, vol. 38, no. 4, pp. 2405–2425, 2022

    work page 2022

  7. [7]

    Centroidal dynamics of a humanoid robot,

    D. E. Orin, A. Goswami, and S.-H. Lee, “Centroidal dynamics of a humanoid robot,”Autonomous robots, vol. 35, no. 2, pp. 161–176, 2013

    work page 2013

  8. [8]

    Dynamic locomotion in the MIT Cheetah 3 through convex model-predictive control,

    J. Di Carlo, P. M. Wensing, B. Katz, G. Bledt, and S. Kim, “Dynamic locomotion in the MIT Cheetah 3 through convex model-predictive control,” inIEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Oct 2018, pp. 1–9

    work page 2018

  9. [9]

    Variational-based optimal control of underactuated balancing for dynamic quadrupeds,

    M. Chignoli and P. M. Wensing, “Variational-based optimal control of underactuated balancing for dynamic quadrupeds,”IEEE Access, vol. 8, pp. 49 785–49 797, 2020

    work page 2020

  10. [10]

    Representation-free model predictive control for dynamic motions in quadrupeds,

    Y . Ding, A. Pandala, C. Li, Y .-H. Shin, and H.-W. Park, “Representation-free model predictive control for dynamic motions in quadrupeds,”IEEE Transactions on Robotics, vol. 37, no. 4, pp. 1154–1171, 2021

    work page 2021

  11. [11]

    Predictive control with indirect adaptive laws for payload transportation by quadrupedal robots,

    L. Amanzadeh, T. Chunawala, R. T. Fawcett, A. Leonessa, and K. Akbari Hamed, “Predictive control with indirect adaptive laws for payload transportation by quadrupedal robots,”IEEE Robotics and Automation Letters, vol. 9, no. 11, pp. 10 359–10 366, 2024

    work page 2024

  12. [12]

    Robust predictive control for quadrupedal locomotion: Learning to close the gap between reduced-and full-order models,

    A. Pandala, R. T. Fawcett, U. Rosolia, A. D. Ames, and K. Ak- bari Hamed, “Robust predictive control for quadrupedal locomotion: Learning to close the gap between reduced-and full-order models,” IEEE Robotics and Automation Letters, vol. 7, no. 3, pp. 6622–6629, 2022

    work page 2022

  13. [13]

    A nonlinear MPC framework for loco-manipulation of quadrupedal robots with non-negligible manipulator dynamics,

    R. S. Sambhus, K. K. Mehta, A. M. Sadeghi, B. M. Imran, J. Kim, T. Chunawala, V . Pastore, S. Vijayan, and K. Akbari Hamed, “A nonlinear MPC framework for loco-manipulation of quadrupedal robots with non-negligible manipulator dynamics,”IEEE Robotics and Automation Letters, vol. 11, no. 4, pp. 4050–4057, 2026

    work page 2026

  14. [14]

    Optimization-based control for dynamic legged robots,

    P. M. Wensing, M. Posa, Y . Hu, A. Escande, N. Mansard, and A. D. Prete, “Optimization-based control for dynamic legged robots,”IEEE Transactions on Robotics, vol. 40, pp. 43–63, 2024

    work page 2024

  15. [15]

    Crocoddyl: An efficient and versatile framework for multi-contact optimal control,

    C. Mastalli, R. Budhiraja, W. Merkt, G. Saurel, B. Hammoud, M. Naveau, J. Carpentier, L. Righetti, S. Vijayakumar, and N. Mansard, “Crocoddyl: An efficient and versatile framework for multi-contact optimal control,” inIEEE International Conference on Robotics and Automation (ICRA), 2020

    work page 2020

  16. [16]

    ProxDDP: Proximal constrained trajectory optimiza- tion,

    W. Jallet, A. Bambade, E. Arlaud, S. El-Kazdadi, N. Mansard, and J. Carpentier, “ProxDDP: Proximal constrained trajectory optimiza- tion,”IEEE Transactions on Robotics, vol. 41, pp. 2605–2624, 2025

    work page 2025

  17. [17]

    Safety barrier certificates for collisions-free multirobot systems,

    L. Wang, A. D. Ames, and M. Egerstedt, “Safety barrier certificates for collisions-free multirobot systems,”IEEE Transactions on Robotics, vol. 33, no. 3, pp. 661–674, 2017

    work page 2017

  18. [18]

    Multirobot adversarial resilience using control barrier functions,

    M. Cavorsi, L. Sabattini, and S. Gil, “Multirobot adversarial resilience using control barrier functions,”IEEE Transactions on Robotics, vol. 40, pp. 797–815, 2024

    work page 2024

  19. [19]

    Walking in narrow spaces: Safety-critical locomotion control for quadrupedal robots with duality-based optimization,

    Q. Liao, Z. Li, A. Thirugnanam, J. Zeng, and K. Sreenath, “Walking in narrow spaces: Safety-critical locomotion control for quadrupedal robots with duality-based optimization,” in2023 IEEE/RSJ Interna- tional Conference on Intelligent Robots and Systems (IROS), 2023, pp. 2723–2730

    work page 2023

  20. [20]

    Multi- layered safety for legged robots via control barrier functions and model predictive control,

    R. Grandia, A. J. Taylor, A. D. Ames, and M. Hutter, “Multi- layered safety for legged robots via control barrier functions and model predictive control,” in2021 IEEE International Conference on Robotics and Automation, 2021, pp. 8352–8358

    work page 2021

  21. [21]

    Safety-critical and distributed nonlinear predictive con- trollers for teams of quadrupedal robots,

    B. M. Imran, J. Kim, T. Chunawala, A. Leonessa, and K. Ak- bari Hamed, “Safety-critical and distributed nonlinear predictive con- trollers for teams of quadrupedal robots,”IEEE Robotics and Automa- tion Letters, vol. 10, no. 9, pp. 9176–9183, 2025

    work page 2025

  22. [22]

    ADMM-based distributed MPC with control barrier functions for safe multi-robot quadrupedal locomotion,

    Y . Zeng, R. S. Sambhus, B. M. Imran, J. Kim, V . Pastore, and K. Akbari Hamed, “ADMM-based distributed MPC with control barrier functions for safe multi-robot quadrupedal locomotion,”arXiv preprint arXiv:2603.19170, 2026

    work page arXiv 2026

  23. [23]

    Layered control for cooperative locomotion of two quadrupedal robots: Centralized and distributed approaches,

    J. Kim, R. T. Fawcett, V . R. Kamidi, A. D. Ames, and K. Ak- bari Hamed, “Layered control for cooperative locomotion of two quadrupedal robots: Centralized and distributed approaches,”IEEE Transactions on Robotics, vol. 39, no. 6, pp. 4728–4748, 2023

    work page 2023

  24. [24]

    Cooperative locomotion via supervi- sory predictive control and distributed nonlinear controllers,

    J. Kim and K. Akbari Hamed, “Cooperative locomotion via supervi- sory predictive control and distributed nonlinear controllers,”Journal of Dynamic Systems, Measurement, and Control, vol. 144, no. 3, p. 031005, Mar. 2022

    work page 2022

  25. [25]

    Distributed data-driven predictive control for multi-agent collaborative legged locomotion,

    R. T. Fawcett, L. Amanzadeh, J. Kim, A. D. Ames, and K. Ak- bari Hamed, “Distributed data-driven predictive control for multi-agent collaborative legged locomotion,” inIEEE International Conference on Robotics and Automation (ICRA), 2023, pp. 9924–9930

    work page 2023

  26. [26]

    Safety-critical coordination for cooperative legged locomotion via control barrier functions,

    J. Kim, J. Lee, and A. D. Ames, “Safety-critical coordination for cooperative legged locomotion via control barrier functions,” in2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2023, pp. 2368–2375

    work page 2023

  27. [27]

    ACLM: ADMM- based distributed model predictive control for collaborative loco- manipulation,

    Z. Zhou, P. Shu, R. Cao, Y . Zhao, and Y . Zhao, “ACLM: ADMM- based distributed model predictive control for collaborative loco- manipulation,”arXiv preprint arXiv:2603.07095, 2026

    work page arXiv 2026

  28. [28]

    Centralized model predictive control for collaborative loco-manipulation,

    F. De Vincenti and S. Coros, “Centralized model predictive control for collaborative loco-manipulation,” inRobotics: Science and Systems (RSS), Daegu, Republic of Korea, 2023

    work page 2023

  29. [29]

    Collabora- tive loco-manipulation for pick-and-place tasks with dynamic reward curriculum,

    T. An, F. De Vincenti, Y . Ma, M. Hutter, and S. Coros, “Collabora- tive loco-manipulation for pick-and-place tasks with dynamic reward curriculum,”arXiv preprint arXiv:2509.13239, 2025

    work page arXiv 2025

  30. [30]

    Discrete-time control barrier function: High-order case and adaptive case,

    Y . Xiong, D.-H. Zhai, M. Tavakoli, and Y . Xia, “Discrete-time control barrier function: High-order case and adaptive case,”IEEE Transac- tions on Cybernetics, vol. 53, no. 5, pp. 3231–3239, 2023

    work page 2023

  31. [31]

    Robust stabilization of periodic gaits for quadrupedal locomotion via QP- based virtual constraint controllers,

    R. T. Fawcett, A. Pandala, A. D. Ames, and K. Akbari Hamed, “Robust stabilization of periodic gaits for quadrupedal locomotion via QP- based virtual constraint controllers,”IEEE Control Systems Letters, pp. 1736–1741, 2021

    work page 2021

  32. [32]

    CasADi – A software framework for nonlinear optimization and optimal control,

    J. A. E. Andersson, J. Gillis, G. Horn, J. B. Rawlings, and M. Diehl, “CasADi – A software framework for nonlinear optimization and optimal control,”Mathematical Programming Computation, vol. 11, no. 1, pp. 1–36, 2019

    work page 2019

  33. [33]

    On the implementation of an interior- point filter line-search algorithm for large-scale nonlinear program- ming,

    A. W ¨achter and L. T. Biegler, “On the implementation of an interior- point filter line-search algorithm for large-scale nonlinear program- ming,”Mathematical programming, vol. 106, pp. 25–57, 2006

    work page 2006

  34. [34]

    Per-contact iteration method for solving contact dynamics,

    J. Hwangbo, J. Lee, and M. Hutter, “Per-contact iteration method for solving contact dynamics,”IEEE Robotics and Automation Letters, vol. 3, no. 2, pp. 895–902, April 2018

    work page 2018

  35. [35]

    Safety-critical centralized nmpc for cooperative payload transporta- tion by two quadrupedal robots. [Online]. Available on YouTube: https://youtu.be/tsV o09eszCo

    “Safety-critical centralized nmpc for cooperative payload transporta- tion by two quadrupedal robots. [Online]. Available on YouTube: https://youtu.be/tsV o09eszCo.”

    work page