Contraction-based Neural Control for Cooperative Aerial Payload Transportation with Variable-length Cables

Hugh H.T. Liu; Longhao Qian; Yi Lok Lo

arxiv: 2606.20127 · v1 · pith:JPBMN2KSnew · submitted 2026-06-18 · 📡 eess.SY · cs.SY

Contraction-based Neural Control for Cooperative Aerial Payload Transportation with Variable-length Cables

Yi Lok Lo , Longhao Qian , Hugh H.T. Liu This is my paper

Pith reviewed 2026-06-26 16:18 UTC · model grok-4.3

classification 📡 eess.SY cs.SY

keywords neural nonlinear controlcontrol contraction metricmulti-drone payload transportvariable-length cablesdecoupled dynamicsobstacle avoidancetrajectory tracking

0 comments

The pith

A neural control contraction metric controller and feedback controller are jointly trained on decoupled dynamics to stabilize the payload subsystem in a multi-drone system with variable-length cables while a separate law varies cable length

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

This paper develops a neural nonlinear control method for multiple drones carrying a rigid payload on cables whose lengths can vary. It first rewrites the equations of motion so that payload motion and cable-length changes are controlled through separate channels. Neural networks are then trained together to make the payload subsystem contract toward desired behaviors. An independent rule adjusts cable lengths to steer around obstacles. Simulations confirm that the combined system follows trajectories and passes through gates.

Core claim

The paper establishes that the multi-drone slung-payload equations can be written in decoupled form so a neural control contraction metric controller and a neural feedback controller can be trained jointly to enforce contraction on the payload subsystem while a separate cable-length control law exploits the extra degree of freedom for obstacle avoidance.

What carries the argument

The jointly trained neural control contraction metric (CCM) controller together with the neural feedback controller on the decoupled payload subsystem.

If this is right

The decoupled structure allows modular controller design on reduced-order subsystems.
Trajectory tracking of the rigid-body payload is achieved under the neural controllers.
Variable cable lengths enable gate traversal in the overall system.
Numerical simulations validate both tracking and avoidance capabilities.

Where Pith is reading between the lines

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

The modular split could be tested on other cooperative systems whose dynamics separate into independent channels.
Hardware experiments would be needed to check whether unmodeled effects break the contraction guarantee obtained in simulation.
The variable-length mechanism suggests direct use in tasks requiring rapid height changes, such as navigating cluttered warehouses.

Load-bearing premise

The equations of motion can be formulated into a decoupled structure in which the payload and cable length dynamics are governed by independent control channels.

What would settle it

A numerical or hardware test in which the payload subsystem fails to contract under the trained controllers even though the assumed decoupling of payload and cable dynamics holds.

read the original abstract

This paper presents a novel neural nonlinear control framework for a multi-drone slung payload system with variable-length cables and a rigid-body payload. The equations of motion are formulated into a decoupled structure, where the payload and cable length dynamics are governed by independent control channels, facilitating modularized controller design on reduced-order subsystems. A neural control contraction metric (CCM) controller and a neural feedback controller are jointly trained to enforce contraction conditions for the payload subsystem. Separately, a cable length control law is derived that exploits the variable-length degree of freedom for obstacle avoidance. Numerical simulations demonstrate trajectory tracking of a rigid-body payload and gate traversal capabilities of the overall system under the proposed control framework.

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.

Desk Editor's Note private letter to a colleague

The paper combines neural CCM training with explicit decoupling for variable-length cable drone payload transport, but the decoupling claim is the part that needs checking first.

read the letter

The core contribution is a modular setup where the multi-drone slung-payload dynamics are split into a payload subsystem handled by a jointly trained neural CCM controller and neural feedback law, plus an independent cable-length law that uses the extra degree of freedom for gate traversal. That split lets them train on reduced-order models rather than the full coupled system, which is a practical move for this kind of cooperative aerial task.

What stands out is the explicit use of variable cable lengths both for control and for obstacle avoidance, something most fixed-length slung-load papers skip. The simulations reportedly show trajectory tracking and gate passing, so the framework at least runs on the claimed model.

The soft spot is the decoupling step itself. The abstract states the equations of motion can be written as two independent channels, but if the full derivation leaves residual coupling through the time-varying attachment points or tension vectors, the separate training no longer guarantees contraction on the real plant. Without seeing the algebra that shows all cross terms cancel or are bounded, it is hard to judge how much approximation is hidden there. The numerical results also lack reported disturbance tests or quantitative error metrics, so the robustness claim stays thin.

This is aimed at researchers already working on contraction-based or learning-based control for aerial robotics. It is not reshaping general theory, but the modular neural-plus-variable-length angle is concrete enough that a serious referee could check the decoupling math and the simulation setup. I would send it to review rather than desk-reject, mainly to get the EOM derivation and any hardware plans on the table.

Referee Report

2 major / 2 minor

Summary. The paper proposes a neural control framework for multi-drone cooperative transport of a rigid payload using variable-length cables. It claims that the nonlinear equations of motion can be exactly decoupled into independent payload and cable-length subsystems, each with its own control channel. A neural CCM controller and neural feedback controller are jointly trained to enforce contraction on the payload subsystem, while a separate analytic law controls cable lengths for obstacle avoidance. Numerical simulations are presented to illustrate trajectory tracking and gate traversal.

Significance. If the decoupling holds rigorously and the contraction conditions are satisfied, the modular neural approach could provide stability guarantees for high-dimensional aerial systems while exploiting variable cable lengths for collision avoidance. The joint training of CCM and feedback terms is a methodological strength that could generalize to other underactuated robotic systems.

major comments (2)

[Abstract / EOM formulation] Abstract and equations-of-motion section: the central claim that the full nonlinear dynamics admit an exact (or rigorously bounded) decoupling into independent payload and cable-length subsystems actuated by separate channels is load-bearing for the entire modular training procedure, yet no explicit algebraic derivation is supplied showing cancellation of all tension-vector and time-varying-attachment cross terms.
[Numerical simulations] Numerical results section: the simulations are stated to demonstrate tracking and gate traversal, but no quantitative metrics (RMS error, contraction rate bounds, disturbance-rejection statistics) or comparison against a coupled baseline are reported, leaving the practical benefit of the decoupled controllers unquantified.

minor comments (2)

[Abstract] Abstract should include at least one quantitative performance figure (e.g., tracking error or success rate) to support the simulation claims.
[Controller design] Notation for the neural CCM metric and the separate cable-length law should be introduced with explicit dependence on the state variables to avoid ambiguity when the subsystems are later recombined.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments on the decoupling claim and simulation evaluation. We address each major comment below.

read point-by-point responses

Referee: [Abstract / EOM formulation] Abstract and equations-of-motion section: the central claim that the full nonlinear dynamics admit an exact (or rigorously bounded) decoupling into independent payload and cable-length subsystems actuated by separate channels is load-bearing for the entire modular training procedure, yet no explicit algebraic derivation is supplied showing cancellation of all tension-vector and time-varying-attachment cross terms.

Authors: We agree that while the manuscript states the EOM admit a decoupled structure, an explicit algebraic derivation showing cancellation of tension-vector and time-varying-attachment cross terms is not supplied. We will add this derivation to the EOM section in the revision. revision: yes
Referee: [Numerical simulations] Numerical results section: the simulations are stated to demonstrate tracking and gate traversal, but no quantitative metrics (RMS error, contraction rate bounds, disturbance-rejection statistics) or comparison against a coupled baseline are reported, leaving the practical benefit of the decoupled controllers unquantified.

Authors: We agree that the current simulations lack quantitative metrics and baseline comparisons. We will add RMS tracking errors, contraction rate bounds, disturbance-rejection statistics, and a comparison to a coupled baseline in the revised numerical results section. revision: yes

Circularity Check

0 steps flagged

No circularity: derivation remains self-contained

full rationale

The paper states that the equations of motion are formulated into a decoupled structure with independent control channels for payload and cable-length subsystems, then trains neural CCM and feedback controllers to enforce contraction on the payload subsystem while deriving a separate cable-length law. No quoted step shows a result reducing to its own inputs by construction, a fitted parameter renamed as prediction, or a load-bearing self-citation chain. The modular training procedure and decoupling claim are presented as algebraic formulation steps whose validity is external to the training outputs themselves.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the modeling choice that dynamics decouple into independent channels; this is a domain assumption rather than a derived result. No free parameters or invented entities are explicitly introduced in the abstract.

axioms (1)

domain assumption The multi-drone slung payload system with variable-length cables admits a decoupled structure separating payload and cable-length dynamics.
Abstract states this formulation facilitates modularized controller design on reduced-order subsystems.

pith-pipeline@v0.9.1-grok · 5644 in / 1211 out tokens · 30779 ms · 2026-06-26T16:18:59.556735+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

23 extracted references · 2 canonical work pages

[1]

Path following control of multiple quadrotors carrying a rigid-body slung payload,

Qian, L., and Liu, H. H., “Path following control of multiple quadrotors carrying a rigid-body slung payload,”AIAA Scitech 2019 Forum, 2019, p. 1172

2019
[2]

Nonlinear control of a multi-drone slung load system,

Al Lawati, M., Zhang, Z., Yan, E., and Lynch, A. F., “Nonlinear control of a multi-drone slung load system,”2025 American Control Conference (ACC), IEEE, 2025, pp. 1989–1996

2025
[3]

Efficient optimization-based cable force allocation for geometric control of a multirotor team transporting a payload,

Wahba, K., and Hönig, W., “Efficient optimization-based cable force allocation for geometric control of a multirotor team transporting a payload,”IEEE Robotics and Automation Letters, Vol. 9, No. 4, 2024, pp. 3688–3695

2024
[4]

Geometriccontrolofmultiplequadrotorstransportingarigid-bodyload

Wu,G.,andSreenath,K.,“Geometriccontrolofmultiplequadrotorstransportingarigid-bodyload.”CDC,2014,pp.6141–6148. 16

2014
[5]

Robust control study for tethered payload transportation using multiple quadrotors,

Qian, L., and Liu, H. H. T., “Robust control study for tethered payload transportation using multiple quadrotors,”Journal of Guidance, Control, and Dynamics, Vol. 45, No. 3, 2022, pp. 434–452

2022
[6]

Geometric control and differential flatness of a quadrotor uav with load suspended from a pulley,

Zeng, J., Kotaru, P., and Sreenath, K., “Geometric control and differential flatness of a quadrotor uav with load suspended from a pulley,”2019 American Control Conference (ACC), IEEE, 2019, pp. 2420–2427

2019
[7]

Anti-swing control and trajectory planning of quadrotor suspended payload system with variable length cable,

Yang, Y., Zhang, D., Xi, H., and Zhang, G., “Anti-swing control and trajectory planning of quadrotor suspended payload system with variable length cable,”Asian Journal of Control, Vol. 24, No. 5, 2022, pp. 2424–2436

2022
[8]

Suppressing UAV payload swing with time-varying cable length through nonlinear coupling,

Huang, J., Tao, H., Wang, Y., and Sun, J.-Q., “Suppressing UAV payload swing with time-varying cable length through nonlinear coupling,”Mechanical Systems and Signal Processing, Vol. 185, 2023, p. 109790. https://doi.org/https://doi.org/10. 1016/j.ymssp.2022.109790, URL https://www.sciencedirect.com/science/article/pii/S0888327022008585

arXiv 2023
[9]

Adaptive trajectory tracking control for the quadrotor aerial transportation system landing a payload onto the mobile platform,

Yu, H., Liang, X., Han, J., and Fang, Y., “Adaptive trajectory tracking control for the quadrotor aerial transportation system landing a payload onto the mobile platform,”IEEE Transactions on Industrial Informatics, Vol. 20, No. 1, 2023, pp. 23–37

2023
[10]

In-flight cable length control for improved quadrotor-based suspended load transportation,

Li, S., Duong, T. T. H., and Zanotto, D., “In-flight cable length control for improved quadrotor-based suspended load transportation,”IEEE Robotics and Automation Letters, Vol. 9, No. 1, 2023, pp. 667–674

2023
[11]

On-boardcableattitudemeasurementandcontrollerforoutdooraerialtransportation,

Prajapati,P.,Parekh,S.,andVashista,V.,“On-boardcableattitudemeasurementandcontrollerforoutdooraerialtransportation,” Robotica, Vol. 40, No. 5, 2022, pp. 1650–1664

2022
[12]

Online Anti-Swing Trajectory Refinement for Variable-Length Cable-Suspended Aerial Transportation Robot,

Yu, H., Yang, Z., He, W., Han, J., Fang, Y., and Liang, X., “Online Anti-Swing Trajectory Refinement for Variable-Length Cable-Suspended Aerial Transportation Robot,”2025 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), IEEE, 2025, pp. 225–231

2025
[13]

Payload trajectory tracking control for aerial transportation systems with cable length online optimization,

Yu, H., Yang, Z., He, W., Han, J., Fang, Y., and Liang, X., “Payload trajectory tracking control for aerial transportation systems with cable length online optimization,”Automatica, Vol. 186, 2026, p. 112864

2026
[14]

Design and control of a variable aerial cable towed system,

Li, Z., Erskine, J., Caro, S., and Chriette, A., “Design and control of a variable aerial cable towed system,”IEEE Robotics and Automation Letters, Vol. 5, No. 2, 2020, pp. 636–643

2020
[15]

Modelpredictivecontrolforamulti-bodyslung-load system,

Tartaglione,G.,D’Amato,E.,Ariola,M.,Rossi,P.S.,andJohansen,T.A.,“Modelpredictivecontrolforamulti-bodyslung-load system,”Robotics and Autonomous Systems, Vol. 92, 2017, pp. 1–11

2017
[16]

Control Contraction Metrics: Convex and Intrinsic Criteria for Nonlinear Feedback Design,

Manchester, I. R., and Slotine, J. J. E., “Control Contraction Metrics: Convex and Intrinsic Criteria for Nonlinear Feedback Design,”IEEE Transactions on Automatic Control, Vol. 62, No. 6, 2017, pp. 3046–3053. https://doi.org/10.1109/TAC.2017. 2668380

work page doi:10.1109/tac.2017 2017
[17]

On contraction analysis for non-linear systems,

Lohmiller, W., and Slotine, J.-J. E., “On contraction analysis for non-linear systems,”Automatica, Vol. 34, No. 6, 1998, pp. 683–696

1998
[18]

Neural contraction metrics for robust estimation and control: A convex optimization approach,

Tsukamoto, H., and Chung, S.-J., “Neural contraction metrics for robust estimation and control: A convex optimization approach,”IEEE Control Systems Letters, Vol. 5, No. 1, 2020, pp. 211–216

2020
[19]

Learning certified control using contraction metric,

Sun, D., Jha, S., and Fan, C., “Learning certified control using contraction metric,”conference on Robot Learning, PMLR, 2021, pp. 1519–1539

2021
[20]

Atheoreticaloverviewofneuralcontractionmetricsforlearning-based control with guaranteed stability,

Tsukamoto, H., Chung, S.-J., Slotine, J.-J., andFan, C., “Atheoreticaloverviewofneuralcontractionmetricsforlearning-based control with guaranteed stability,”2021 60th IEEE Conference on Decision and Control (CDC), IEEE, 2021, pp. 2949–2954

2021
[21]

Neural Robust Control on Lie Groups Using Contraction Methods (Extended Version),

Lo, Y. L., Qian, L., and Liu, H. H. T., “Neural Robust Control on Lie Groups Using Contraction Methods (Extended Version),” arXiv preprint arXiv:2604.01448, 2026

arXiv 2026
[22]

Control contraction metrics on submanifolds,

Wu, D., Yi, B., and Manchester, I. R., “Control contraction metrics on submanifolds,”2024 IEEE 63rd Conference on Decision and Control (CDC), IEEE, 2024, pp. 3735–3740

2024
[23]

Naomi Ehrich Leonard

Lee, T., Leok, M., and McClamroch, N. H., “Geometric tracking control of a quadrotor UAV on SE(3),”Proceedings of the IEEE Conference on Decision and Control, 2010, pp. 5420–5425. https://doi.org/10.1109/CDC.2010.5717652. 17

work page doi:10.1109/cdc.2010.5717652 2010

[1] [1]

Path following control of multiple quadrotors carrying a rigid-body slung payload,

Qian, L., and Liu, H. H., “Path following control of multiple quadrotors carrying a rigid-body slung payload,”AIAA Scitech 2019 Forum, 2019, p. 1172

2019

[2] [2]

Nonlinear control of a multi-drone slung load system,

Al Lawati, M., Zhang, Z., Yan, E., and Lynch, A. F., “Nonlinear control of a multi-drone slung load system,”2025 American Control Conference (ACC), IEEE, 2025, pp. 1989–1996

2025

[3] [3]

Efficient optimization-based cable force allocation for geometric control of a multirotor team transporting a payload,

Wahba, K., and Hönig, W., “Efficient optimization-based cable force allocation for geometric control of a multirotor team transporting a payload,”IEEE Robotics and Automation Letters, Vol. 9, No. 4, 2024, pp. 3688–3695

2024

[4] [4]

Geometriccontrolofmultiplequadrotorstransportingarigid-bodyload

Wu,G.,andSreenath,K.,“Geometriccontrolofmultiplequadrotorstransportingarigid-bodyload.”CDC,2014,pp.6141–6148. 16

2014

[5] [5]

Robust control study for tethered payload transportation using multiple quadrotors,

Qian, L., and Liu, H. H. T., “Robust control study for tethered payload transportation using multiple quadrotors,”Journal of Guidance, Control, and Dynamics, Vol. 45, No. 3, 2022, pp. 434–452

2022

[6] [6]

Geometric control and differential flatness of a quadrotor uav with load suspended from a pulley,

Zeng, J., Kotaru, P., and Sreenath, K., “Geometric control and differential flatness of a quadrotor uav with load suspended from a pulley,”2019 American Control Conference (ACC), IEEE, 2019, pp. 2420–2427

2019

[7] [7]

Anti-swing control and trajectory planning of quadrotor suspended payload system with variable length cable,

Yang, Y., Zhang, D., Xi, H., and Zhang, G., “Anti-swing control and trajectory planning of quadrotor suspended payload system with variable length cable,”Asian Journal of Control, Vol. 24, No. 5, 2022, pp. 2424–2436

2022

[8] [8]

Suppressing UAV payload swing with time-varying cable length through nonlinear coupling,

Huang, J., Tao, H., Wang, Y., and Sun, J.-Q., “Suppressing UAV payload swing with time-varying cable length through nonlinear coupling,”Mechanical Systems and Signal Processing, Vol. 185, 2023, p. 109790. https://doi.org/https://doi.org/10. 1016/j.ymssp.2022.109790, URL https://www.sciencedirect.com/science/article/pii/S0888327022008585

arXiv 2023

[9] [9]

Adaptive trajectory tracking control for the quadrotor aerial transportation system landing a payload onto the mobile platform,

Yu, H., Liang, X., Han, J., and Fang, Y., “Adaptive trajectory tracking control for the quadrotor aerial transportation system landing a payload onto the mobile platform,”IEEE Transactions on Industrial Informatics, Vol. 20, No. 1, 2023, pp. 23–37

2023

[10] [10]

In-flight cable length control for improved quadrotor-based suspended load transportation,

Li, S., Duong, T. T. H., and Zanotto, D., “In-flight cable length control for improved quadrotor-based suspended load transportation,”IEEE Robotics and Automation Letters, Vol. 9, No. 1, 2023, pp. 667–674

2023

[11] [11]

On-boardcableattitudemeasurementandcontrollerforoutdooraerialtransportation,

Prajapati,P.,Parekh,S.,andVashista,V.,“On-boardcableattitudemeasurementandcontrollerforoutdooraerialtransportation,” Robotica, Vol. 40, No. 5, 2022, pp. 1650–1664

2022

[12] [12]

Online Anti-Swing Trajectory Refinement for Variable-Length Cable-Suspended Aerial Transportation Robot,

Yu, H., Yang, Z., He, W., Han, J., Fang, Y., and Liang, X., “Online Anti-Swing Trajectory Refinement for Variable-Length Cable-Suspended Aerial Transportation Robot,”2025 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), IEEE, 2025, pp. 225–231

2025

[13] [13]

Payload trajectory tracking control for aerial transportation systems with cable length online optimization,

Yu, H., Yang, Z., He, W., Han, J., Fang, Y., and Liang, X., “Payload trajectory tracking control for aerial transportation systems with cable length online optimization,”Automatica, Vol. 186, 2026, p. 112864

2026

[14] [14]

Design and control of a variable aerial cable towed system,

Li, Z., Erskine, J., Caro, S., and Chriette, A., “Design and control of a variable aerial cable towed system,”IEEE Robotics and Automation Letters, Vol. 5, No. 2, 2020, pp. 636–643

2020

[15] [15]

Modelpredictivecontrolforamulti-bodyslung-load system,

Tartaglione,G.,D’Amato,E.,Ariola,M.,Rossi,P.S.,andJohansen,T.A.,“Modelpredictivecontrolforamulti-bodyslung-load system,”Robotics and Autonomous Systems, Vol. 92, 2017, pp. 1–11

2017

[16] [16]

Control Contraction Metrics: Convex and Intrinsic Criteria for Nonlinear Feedback Design,

Manchester, I. R., and Slotine, J. J. E., “Control Contraction Metrics: Convex and Intrinsic Criteria for Nonlinear Feedback Design,”IEEE Transactions on Automatic Control, Vol. 62, No. 6, 2017, pp. 3046–3053. https://doi.org/10.1109/TAC.2017. 2668380

work page doi:10.1109/tac.2017 2017

[17] [17]

On contraction analysis for non-linear systems,

Lohmiller, W., and Slotine, J.-J. E., “On contraction analysis for non-linear systems,”Automatica, Vol. 34, No. 6, 1998, pp. 683–696

1998

[18] [18]

Neural contraction metrics for robust estimation and control: A convex optimization approach,

Tsukamoto, H., and Chung, S.-J., “Neural contraction metrics for robust estimation and control: A convex optimization approach,”IEEE Control Systems Letters, Vol. 5, No. 1, 2020, pp. 211–216

2020

[19] [19]

Learning certified control using contraction metric,

Sun, D., Jha, S., and Fan, C., “Learning certified control using contraction metric,”conference on Robot Learning, PMLR, 2021, pp. 1519–1539

2021

[20] [20]

Atheoreticaloverviewofneuralcontractionmetricsforlearning-based control with guaranteed stability,

Tsukamoto, H., Chung, S.-J., Slotine, J.-J., andFan, C., “Atheoreticaloverviewofneuralcontractionmetricsforlearning-based control with guaranteed stability,”2021 60th IEEE Conference on Decision and Control (CDC), IEEE, 2021, pp. 2949–2954

2021

[21] [21]

Neural Robust Control on Lie Groups Using Contraction Methods (Extended Version),

Lo, Y. L., Qian, L., and Liu, H. H. T., “Neural Robust Control on Lie Groups Using Contraction Methods (Extended Version),” arXiv preprint arXiv:2604.01448, 2026

arXiv 2026

[22] [22]

Control contraction metrics on submanifolds,

Wu, D., Yi, B., and Manchester, I. R., “Control contraction metrics on submanifolds,”2024 IEEE 63rd Conference on Decision and Control (CDC), IEEE, 2024, pp. 3735–3740

2024

[23] [23]

Naomi Ehrich Leonard

Lee, T., Leok, M., and McClamroch, N. H., “Geometric tracking control of a quadrotor UAV on SE(3),”Proceedings of the IEEE Conference on Decision and Control, 2010, pp. 5420–5425. https://doi.org/10.1109/CDC.2010.5717652. 17

work page doi:10.1109/cdc.2010.5717652 2010