Pith. sign in

REVIEW 2 major objections 5 minor 43 references

A single neural policy trained in a 2D simulation transfers zero-shot to real 3D drone teams carrying payloads, and scales to team sizes beyond its training range.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · deepseek-v4-flash

2026-08-01 09:41 UTC pith:EOVVHXJ6

load-bearing objection The hardware results are a real step forward, but the paper's formal safety guarantee does not cover the policy it evaluates, and the 'fully distributed' claim is softened by centralized emulation. the 2 major comments →

arxiv 2607.20665 v1 pith:EOVVHXJ6 submitted 2026-07-22 cs.RO cs.MAcs.SYeess.SY

Safe and Scalable Multi-Drone Payload Transport via CBF-based Reinforcement Learning with Zero-Shot Sim-to-Real Transfer

Jaeyoun Choi , Oswin So , Songyuan Zhang , Cooper Taylor , Chuchu Fan This is my paper
classification cs.RO cs.MAcs.SYeess.SY
keywords multi-drone payload transportcable-suspended loadcontrol barrier functionsmulti-agent reinforcement learningzero-shot sim-to-realdistributed policydomain randomizationsafety certificates
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

This paper is trying to establish that safe, scalable, cable-suspended multi-drone payload transport can be achieved by a single, fully distributed neural policy trained in a minimal 2D simulation and deployed on real 3D quadrotors without any fine-tuning. The authors report that this policy, trained only on teams of three to five drones, generalizes zero-shot to six drones and to multi-team scenes in which other teams act as dynamic obstacles. The proposed mechanism is a 2D abstraction that preserves the drone-payload force coupling, combined with a graph control barrier function whose constraints are tightened by a worst-case tracking error to bridge the discrete-to-continuous gap. A sympathetic reader would care because, if correct, it takes a step toward practical deployment of aerial manipulation systems in dynamic, unstructured environments where centralized planners and high-fidelity models are impractical.

Core claim

The paper's central claim is that cable-suspended multi-drone payload transport does not need a high-fidelity 3D model or a centralized planner to be safe and scalable. Instead, a single fully distributed neural policy, trained in a minimal 2D abstraction with random team sizes (3–5) and cable stiffness, transfers zero-shot to real quadrotor teams of up to six drones and to multi-group scenes where other teams are perceived as moving LiDAR obstacles. The paper reports that all 3-drone Easy and Hard scenarios succeed, that the policy generalizes to 4, 5, and 6 drones even though only 3–5 were seen in training, and that two independent groups can fly past each other without collision. The acco

What carries the argument

The load-bearing mechanism is a three-part stack. First, a minimal 2D abstraction models each drone as a point mass connected to the payload by a virtual spring, preserving the net force and torque that the drone configuration exerts on the payload while discarding out-of-plane dynamics; this makes large-scale reinforcement learning tractable. Second, DGPPO (Discrete Graph Control Barrier Function Proximal Policy Optimization) jointly learns a shared distributed policy and a discrete-time graph control barrier function—a safety certificate computed from local observations—so each agent can judge safety without global state. Third, a hierarchical execution bridge tightens each safety constrai

Load-bearing premise

The safety theorem assumes a known uniform tracking-error bound epsilon_trk and assumes the learned policy keeps the tightened DGCBF constraints forward-invariant at every decision step; the paper measures only mean tracking error and never verifies the forward-invariance of the neural policy, so both assumptions are unproven in hardware.

What would settle it

On the real platform, run the deployment pipeline over a long sequence of aggressive maneuvers, logging each drone's position tracking error at full rate; if the observed maximum exceeds the epsilon_trk used to tighten the constraints (with the same safety margins), the premise of Theorem 1 fails. Separately, log the learned policy's constraint value h_tight at each 10 Hz decision step and search for any step where h_tight(ok) <= 0 but h_tight(ok+1) > 0, which directly falsifies Eq. (11) and voids the continuous-time safety claim.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • A low-fidelity planar model with virtual-spring tensions is sufficient to learn real-world cooperative transport behaviors; teams can be scaled by training once and reusing the same policy rather than retraining per team size.
  • Because safety is certified in observation space rather than a global state, the approach extends to multi-group settings where other robots appear only as LiDAR points, without any explicit interaction model.
  • The hierarchical two-timescale architecture (10 Hz decision layer, high-rate tracking) with constraint tightening gives a template for porting other discrete-time learned safety certificates to continuous-time hardware.
  • In simulations the policy maintains high safety rates up to 8 agents (beyond the 3–5 training range), suggesting the learned coordination does not rely on a fixed formation count.
  • The one incomplete 6-drone hardware trial—where the payload plateaus near the goal with a final error of 0.5 m—is described by the authors as a terminal pose-refinement issue from reward shaping, not a safety violation, and points to a specific improvement target.

Where Pith is reading between the lines

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

  • Because the paper only reports mean tracking error per trial while the guarantee needs a uniform worst-case bound, an inference is that the safety margin should be validated against true worst-case error during aggressive maneuvers; if worst-case error exceeds the assumed epsilon_trk, the tightening is insufficient.
  • The same local LiDAR-based observation could support a fleet-wide shared policy in which independent teams or even heterogeneous agents coordinate implicitly, without the paper claiming or testing this.
  • The reward-shaping plateau seen in the 6-drone hard trial suggests a practical fix: add a goal-approach switching term or dead-zone reward; the paper attributes the failure to reward shaping, so testing that fix would separate a reward design issue from a safety or scalability issue.
  • Since the 2D abstraction assumes regulated altitude and planar payload motion, a natural inference is that the approach could be extended to 3D by stacking a second planar abstraction with modified spring-tension coupling, but cable swing and out-of-plane forces would need explicit accounting.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

2 major / 5 minor

Summary. The paper proposes a learning-based framework for safe, scalable, multi-drone cable-suspended payload transport. The authors introduce a minimal 2D planar abstraction of the coupled drone-payload dynamics, train a single distributed policy with domain randomization via Discrete Graph Control Barrier Function Proximal Policy Optimization (DGPPO), and present a continuous-time safety theorem that bridges discrete-time DGCBF constraints to real-world execution under bounded tracking error. The experimental section reports simulation comparisons against other safe MARL baselines and hardware experiments on Crazyflie teams of 3–6 drones, including multi-group scenarios in which other teams act as dynamic obstacles.

Significance. If the safety theorem could be instantiated for the trained policy, the paper would be a valuable step toward formal guarantees in learning-based multi-robot systems. The hardware results are also notable: a single learned policy transfers zero-shot to teams larger than those seen in training and to dynamic multi-group obstacle environments. However, the central formal claim is not established. The paper's own simulation data report a nominal safety rate of 76.2%, which directly contradicts the forward-invariance hypothesis required by Theorem 1. The tracking-error bound and Lipschitz constants needed to apply the theorem are never quantified. Thus the 'safe' guarantee in the title and abstract is unsupported, although the empirical framework itself remains potentially useful if the claims are appropriately revised.

major comments (2)
  1. [§VI.A.5 / Theorem 1 / Eq. (11)] The paper's own evaluation falsifies the key hypothesis of Theorem 1. In §VI.A.5 the nominal safety rate is reported as 76.2%, where safety rate is defined as the fraction of agents satisfying all safety constraints throughout execution (§VI.A.2). A rate below 100% means that for some agent, episode, and time step, h_i^{(m)} > 0 for some m. Since Eq. (10) defines h_tight = h + L_{i,m} L_{O,i} ε_trk with nonnegative constants, h_i^{(m)} > 0 implies h_{i,tight}^{(m)} > 0. Hence the tightened constraints are not forward invariant at decision steps, contradicting Eq. (11). Consequently Theorem 1 does not apply to the trained policy. This is load-bearing because 'safe' appears in the title, abstract, and contribution list. The authors must either verify Eq. (11) (e.g., by formal or statistical certification) or explicitly downgrade the guarantee to an empirical statement.
  2. [§V, Assumption 1 and Eqs. (7)–(10)] Even if the simulation contradiction were resolved, the theorem is not instantiated for the real system. Assumption 1 requires a known uniform bound ε_trk such that ||p_i(t) − p_{i,d}(t)|| ≤ ε_trk for all agents and all t. The only hardware evidence is a mean tracking error below 5.8 cm (§VI.B), which does not provide a uniform max-norm bound. Moreover, the Lipschitz constants L_{i,m} (Assumption 2) and L_{O,i} (Assumption 3) are never computed for the actual observation features, so the tightened constraints (10) cannot be evaluated in practice. Assumptions 2–3 are also questionable for LiDAR observations, which can change discontinuously under occlusion. The paper should either provide verified bounds for these quantities or clearly state that Theorem 1 is a purely conditional statement with no instantiated application.
minor comments (5)
  1. [Eq. (18)] The displayed h_i is a vector but the text and theorem treat h^{(m)} as scalar constraint functions. Please define each component explicitly. Also, the second component appears to be missing a bracket or exponent: `(r − min_j ||p_i − o_k||)^2 d_po` is the likely intended form; the current typesetting is ambiguous.
  2. [Fig. 1(b) and §IV.B] The term 'safety-certified desired waypoints' in Fig. 1(b) is not supported by any verification result. DGPPO trains with constraint filtering/penalties; 'certified' should be reserved for a policy that provably satisfies Eq. (11).
  3. [§VI.B and §VI.C] The acceleration scaling factor of 15 and the increased communication radius of 0.30 m in the multi-group experiment are introduced without justification. These are additional free parameters not covered by domain randomization; their effect on the sim-to-real gap should be discussed.
  4. [§VI.B] Hardware experiments use a centralized computer for state estimation, with 'decentralized execution emulated.' The phrase 'fully distributed execution' in the conclusion overstates the setup; the policy is distributed, but perception and state estimation are not.
  5. [§VI.A.4 vs §VI.A.5] Fig. 2 suggests high safety rates for DGPPO, while §VI.A.5 reports a nominal safety rate of 76.2% under the robustness evaluation. These numbers need reconciliation; the discrepancy in nominal conditions makes the safety-rate metric difficult to interpret.

Circularity Check

0 steps flagged

No significant circularity: the safety theorem is a conditional sufficient-condition result, and the empirical benchmarks are external.

full rationale

The paper's central formal claim is Theorem 1, which derives continuous-time safety from three stated assumptions plus the discrete-time forward-invariance condition (11). This is not a circular reduction: (11) is a stronger condition than the conclusion (12), and the proof is a standard Lipschitz/epsilon-tightening argument rather than an appeal to the conclusion. The paper does not verify (11) for the learned policy, and its reported 76.2% nominal safety rate in Section VI.A.5 is a serious soundness concern, but an unverified or even empirically contradicted hypothesis is a verification gap, not a definitional equivalence. The DGPPO and DGCBF machinery is taken from the authors' prior work [37], [38]; this is self-citation, but those are separate peer-reviewed publications with their own stated assumptions, and the present paper does not invoke a uniqueness theorem or forbid alternatives through them. The hardware experiments, including the multi-group dynamic-obstacle runs, are external benchmarks conducted after training and do not reduce to the paper's own fitted values. The 15x acceleration scaling and the 5.8 cm mean tracking error are explicitly reported as empirical observations, not disguised as predictions. No equation in the paper is defined in terms of the quantity it purports to derive, and no fitted parameter is renamed as a prediction. Therefore, there is no significant circularity.

Axiom & Free-Parameter Ledger

7 free parameters · 8 axioms · 1 invented entities

The central safety claim rests on the 2D virtual-spring abstraction, the imported DGCBF guarantee from prior work, and two unverified hypotheses: a known worst-case tracking bound and forward invariance of tightened constraints by a learned neural policy. Many task-shaping constants (reward weights, margins, 15x acceleration scaling) are hand-set, and the safety constraint distances include unspecified values. This is a high burden of unverified support for the word 'safe'.

free parameters (7)
  • K (virtual spring stiffness) = sampled uniformly from [0.05, 0.30] N/m in training
    Defines the tension model T_i = K(...) in Eq. (5); randomized rather than fit, but a modeling choice central to the 2D abstraction.
  • Acceleration scaling factor = 15
    Applied to commanded accelerations on hardware; 'selected empirically' in Section VI.B to stabilize closed-loop behavior.
  • Reward weights (w_d, w_theta, w_thr, w_u, w_s) = 0.06, 0.06, 0.001, 0.1, 0.1
    Hand-chosen in Eq. (17) to shape task behavior; not derived.
  • Safety margin epsilon and clipping = epsilon = 0.5, clip [-1, 1]
    Margin-shifted safety constraints in Eq. (18); selected by hand and not tied to the tracking-error bound in Theorem 1.
  • Tracking error bound epsilon_trk = mean < 5.8 cm (claimed)
    Assumption 1 requires a known uniform bound; hardware section reports a mean, so this is effectively an empirical value, not a certified worst-case bound.
  • Agent attachment tolerance d_tol / payload-obstacle distance d_po = not specified
    Appear in Eq. (18) safety constraints but their numerical values are not given.
  • Communication and LiDAR radii = R = 0.25 m (0.30 m in multi-group), LiDAR range 0.5 m
    Sensing parameters used in observations; the multi-group experiment modifies R, so behavior depends on hand-set values.
axioms (8)
  • domain assumption 2D abstraction assumptions: altitude regulated, tension proportional to projected cable length, payload nearly planar, fixed symmetric attachment points
    Used to derive Eq. (4)-(5) in Section IV.A; if these fail, the simulator does not model real cable-suspended dynamics.
  • domain assumption Virtual spring tension model Eq. (5)
    Replaces cable forces with K times the planar offset; a modeling assumption not derived from first principles.
  • domain assumption Double-integrator discretization with dt = 0.1 s
    Decision-level dynamics used for training and safety; discretization error is not analyzed beyond five substeps.
  • standard math Corollary 1 (DGCBF from [37])
    Imported without proof from the authors' own ICLR 2025 paper; the safety certificate relies on this result.
  • domain assumption Assumption 1: known uniform tracking-error bound epsilon_trk
    Theorem 1 needs a uniform worst-case bound; only a mean tracking error is reported, so this is an unverified premise.
  • domain assumption Assumption 2: safety constraints are Lipschitz
    Needed for the epsilon-tightening argument; Lipschitz constants are not computed.
  • domain assumption Assumption 3: observation error bounded by tracking error via L_O,i
    Binds tracking error to observation error; the constant L_O,i is not quantified.
  • ad hoc to paper Eq. (11): learned policy renders tightened constraints forward invariant
    The theorem's key hypothesis; the paper does not verify that the learned neural policy satisfies this invariant.
invented entities (1)
  • Planar virtual-spring cable abstraction no independent evidence
    purpose: Represents cable-suspended payload dynamics in 2D for scalable RL training (Section IV.A)
    No direct force/tension measurements validate the spring model; only indirect sim-to-real success is offered.

pith-pipeline@v1.3.0-alltime-deepseek · 12282 in / 15640 out tokens · 119943 ms · 2026-08-01T09:41:46.269364+00:00 · methodology

0 comments
read the original abstract

Multi-drone payload transportation has emerged as a promising research paradigm with potential applications in construction, logistics, and disaster response. However, the complex coupled dynamics among drones, cables, and payloads pose significant challenges, and existing approaches remain limited in safety and scalability, particularly in dynamic and unstructured environments. In this work, we propose a learning-based framework for safe and scalable multi-drone cooperative payload transport. We introduce a minimal 2D abstraction that preserves the task-relevant drone-payload coupling required for coordination and safety, while remaining computationally efficient for large-scale learning. Using domain randomization over team size and physical parameters, we train a fully distributed policy via Discrete Graph Control Barrier Function Proximal Policy Optimization (DGPPO), enabling robust zero-shot sim-to-real transfer without fine-tuning. Extensive real-world evaluations demonstrate that a single learned policy generalizes across varying team sizes and task scenarios. Furthermore, multi-group hardware experiments show that the same policy can safely operate in dynamic environments, where other drone teams act as moving obstacles. These results indicate that the proposed framework enables efficient, safe, and scalable multi-drone payload transportation with strong generalization to complex real-world conditions.

Figures

Figures reproduced from arXiv: 2607.20665 by Chuchu Fan, Cooper Taylor, Jaeyoun Choi, Oswin So, Songyuan Zhang.

Figure 1
Figure 1. Figure 1: Distributed learning framework for safe and scalable multi-drone transportation. (a) Real-world cable-suspended transport and corresponding 2D abstraction. Drones are modeled as point masses connected to the payload via virtual springs, and obstacles are detected with LiDAR. (b) Control architecture and discrete-to-continuous safety bridge. A distributed policy π generates DGCBF-certified waypoints from lo… view at source ↗
Figure 2
Figure 2. Figure 2: Simulation results across different team sizes. For each number of agents (n ∈ {3, 4, 5, 6, 8}), we evaluate poli￾cies on 25 randomly generated scenarios. Each marker reports the mean reward and safety rate across five evaluation seeds (125 scenarios total), with error bars showing variability across seeds. We compare several distributed policy optimization backbones under the same setup. Policies were tra… view at source ↗
Figure 3
Figure 3. Figure 3: Hardware validation of distributed multi-drone transportation. (a) Easy scenarios where obstacles do not significantly affect the payload trajectory. (b) Hard scenarios where obstacles actively constrain the transport trajectories. (c) Scalability and generalization of a single learned policy evaluated with 3–6 agents. While DGPPO and the β = 0.5 penalty variant are similar in aggregate, under a stricter 0… view at source ↗
Figure 4
Figure 4. Figure 4: Multi-group multi-drone transportation experiment. Time-lapse snapshots and recorded trajectories show two independent teams transporting separate payloads while perceiving the other team as dynamic obstacles via LiDAR. To validate policy performance, we design 10 hardware test scenarios, comprising 5 Easy scenarios (Fig. 3a) and 5 Hard scenarios (Fig. 3b). In each scenario, the initial positions of the dr… view at source ↗

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

43 extracted references · 1 linked inside Pith

  1. [1]

    Cooperative transport by ants and robots,

    C. R. Kube and E. Bonabeau, “Cooperative transport by ants and robots,” Robotics and autonomous systems, vol. 30, no. 1-2, pp. 85–101, 2000. IEEE ROBOTICS AND AUTOMATION LETTERS. PREPRINT VERSION. ACCEPTED JULY , 2026 8

    2000

  2. [2]

    Cooperative object transport in multi-robot systems: A review of the state-of-the-art,

    E. Tuci, M. H. Alkilabi, and O. Akanyeti, “Cooperative object transport in multi-robot systems: A review of the state-of-the-art,”Frontiers in Robotics and AI, vol. 5, p. 59, 2018

    2018

  3. [3]

    Multi-robot systems and cooperative object transport: Communications, platforms, and challenges,

    X. An, C. Wu, Y . Lin, M. Lin, T. Yoshinaga, and Y . Ji, “Multi-robot systems and cooperative object transport: Communications, platforms, and challenges,”IEEE Open Journal of the Computer Society, vol. 4, pp. 23–36, 2023

    2023

  4. [4]

    A mechanical route for cooperative transport in autonomous robotic swarms,

    E. Arbel, L. Buise, C. van Waes, N. Oppenheimer, Y . Lahini, and M. Y . Ben Zion, “A mechanical route for cooperative transport in autonomous robotic swarms,”Nature Communications, vol. 16, no. 1, p. 7519, 2025

    2025

  5. [5]

    Multi-robot formation control and object transport in dynamic environments via constrained optimiza- tion,

    J. Alonso-Mora, S. Baker, and D. Rus, “Multi-robot formation control and object transport in dynamic environments via constrained optimiza- tion,”The International Journal of Robotics Research, vol. 36, no. 9, pp. 1000–1021, 2017

    2017

  6. [6]

    Two-layers workspace: A new approach to cooperative object transportation with obstacle avoidance for multi-robot system,

    S. K. A. De Sousa, R. C. S. Freire, E. A. N. Carvalho, L. Molina, P. C. Santos, and E. O. Freire, “Two-layers workspace: A new approach to cooperative object transportation with obstacle avoidance for multi-robot system,”IEEE Access, vol. 10, pp. 6929–6939, 2022

    2022

  7. [7]

    A review of real- time implementable cooperative aerial manipulation systems,

    S. C. Barakou, C. S. Tzafestas, and K. P. Valavanis, “A review of real- time implementable cooperative aerial manipulation systems,”Drones, vol. 8, no. 5, p. 196, 2024

    2024

  8. [8]

    Scalable cooperative transport of cable-suspended loads with uavs using distributed trajectory optimization,

    B. E. Jackson, T. A. Howell, K. Shah, M. Schwager, and Z. Manchester, “Scalable cooperative transport of cable-suspended loads with uavs using distributed trajectory optimization,”IEEE Robotics and Automation Letters, vol. 5, no. 2, pp. 3368–3374, 2020

    2020

  9. [9]

    Nonlinear model predictive control for cooper- ative transportation and manipulation of cable suspended payloads with multiple quadrotors,

    G. Li and G. Loianno, “Nonlinear model predictive control for cooper- ative transportation and manipulation of cable suspended payloads with multiple quadrotors,” in2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2023, pp. 5034–5041

    2023

  10. [10]

    Planning and control for collision- free cooperative aerial transportation,

    H. Lee, H. Kim, and H. J. Kim, “Planning and control for collision- free cooperative aerial transportation,”IEEE Transactions on Automation Science and Engineering, vol. 15, no. 1, pp. 189–201, 2016

    2016

  11. [11]

    Distributed model predictive control for uavs collaborative payload transport,

    J. Wehbeh, S. Rahman, and I. Sharf, “Distributed model predictive control for uavs collaborative payload transport,” in2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2020, pp. 11 666–11 672

    2020

  12. [12]

    Dynamic collaboration without communication: Vision-based cable-suspended load transport with two quadrotors,

    M. Gassner, T. Cieslewski, and D. Scaramuzza, “Dynamic collaboration without communication: Vision-based cable-suspended load transport with two quadrotors,” in2017 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2017, pp. 5196–5202

    2017

  13. [13]

    Hpa-mpc: Hybrid perception- aware nonlinear model predictive control for quadrotors with suspended loads,

    M. Sarvaiya, G. Li, and G. Loianno, “Hpa-mpc: Hybrid perception- aware nonlinear model predictive control for quadrotors with suspended loads,”IEEE Robotics and Automation Letters, vol. 10, no. 1, pp. 358– 365, 2025

    2025

  14. [14]

    Human-aware physical human–robot collaborative transportation and manipulation with multiple aerial robots,

    G. Li, X. Liu, and G. Loianno, “Human-aware physical human–robot collaborative transportation and manipulation with multiple aerial robots,”IEEE Transactions on Robotics, vol. 41, pp. 762–781, 2025

    2025

  15. [15]

    A novel robotic platform for aerial manipulation using quadrotors as rotating thrust generators,

    H.-N. Nguyen, S. Park, J. Park, and D. Lee, “A novel robotic platform for aerial manipulation using quadrotors as rotating thrust generators,” IEEE Transactions on Robotics, vol. 34, no. 2, pp. 353–369, 2018

    2018

  16. [16]

    Cooperative manipulation and transportation with aerial robots,

    N. Michael, J. Fink, and V . Kumar, “Cooperative manipulation and transportation with aerial robots,”Autonomous Robots, vol. 30, no. 1, pp. 73–86, 2011

    2011

  17. [17]

    Load transportation system based on autonomous small size helicopters,

    M. Bernard, K. Kondak, and G. Hommel, “Load transportation system based on autonomous small size helicopters,”The aeronautical journal, vol. 114, no. 1153, pp. 191–198, 2010

    2010

  18. [18]

    Full- pose manipulation control of a cable-suspended load with multiple uavs under uncertainties,

    D. Sanalitro, H. J. Savino, M. Tognon, J. Cort ´es, and A. Franchi, “Full- pose manipulation control of a cable-suspended load with multiple uavs under uncertainties,”IEEE Robotics and Automation Letters, vol. 5, no. 2, pp. 2185–2191, 2020

    2020

  19. [19]

    Nonlinear mpc for full-pose manipulation of a cable-suspended load using multiple uavs,

    S. Sun and A. Franchi, “Nonlinear mpc for full-pose manipulation of a cable-suspended load using multiple uavs,” in2023 International Conference on Unmanned Aircraft Systems (ICUAS). IEEE, 2023, pp. 969–975

    2023

  20. [20]

    Optimal trajectory planning for cooperative manipulation with multiple quadrotors using control barrier functions,

    A. Pallar, G. Li, M. Sarvaiya, and G. Loianno, “Optimal trajectory planning for cooperative manipulation with multiple quadrotors using control barrier functions,” in2025 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2025, pp. 2808–2814

    2025

  21. [21]

    Agile and cooperative aerial manipulation of a cable-suspended load,

    S. Sun, X. Wang, D. Sanalitro, A. Franchi, M. Tognon, and J. Alonso- Mora, “Agile and cooperative aerial manipulation of a cable-suspended load,”Science Robotics, vol. 10, no. 107, p. eadu8015, 2025

    2025

  22. [22]

    Decentralized model predictive control for equilibrium-based collaborative uav bar transporta- tion,

    R. C. Sundin, P. Roque, and D. V . Dimarogonas, “Decentralized model predictive control for equilibrium-based collaborative uav bar transporta- tion,” in2022 International Conference on Robotics and Automation (ICRA). IEEE, 2022, pp. 4915–4921

    2022

  23. [23]

    Cooperative transportation using multiple single-rotor robots and decentralized control for unknown pay- loads,

    K. Oishi, Y . Amano, and T. Jimbo, “Cooperative transportation using multiple single-rotor robots and decentralized control for unknown pay- loads,” in2022 International Conference on Robotics and Automation (ICRA). IEEE, 2022, pp. 2024–2030

    2022

  24. [24]

    Decen- tralized motion control in a cabled-based multi-drone load transport system,

    K. Mohammadi, M. Jafarinasab, S. Sirouspour, and E. Dyer, “Decen- tralized motion control in a cabled-based multi-drone load transport system,” in2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2018, pp. 4198–4203

    2018

  25. [25]

    Cooperative transportation of cable suspended payloads with mavs using monocular vision and inertial sensing,

    G. Li, R. Ge, and G. Loianno, “Cooperative transportation of cable suspended payloads with mavs using monocular vision and inertial sensing,”IEEE Robotics and Automation Letters, vol. 6, no. 3, pp. 5316– 5323, 2021

    2021

  26. [26]

    Collabo- rative transportation using mavs via passive force control,

    A. Tagliabue, M. Kamel, S. Verling, R. Siegwart, and J. Nieto, “Collabo- rative transportation using mavs via passive force control,” in2017 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2017, pp. 5766–5773

    2017

  27. [27]

    Geometric control of quadrotor uavs transporting a cable- suspended rigid body,

    T. Lee, “Geometric control of quadrotor uavs transporting a cable- suspended rigid body,”IEEE Transactions on Control Systems Tech- nology, vol. 26, no. 1, pp. 255–264, 2017

    2017

  28. [28]

    Geometric control of multiple quadrotors transporting a rigid-body load,

    G. Wu and K. Sreenath, “Geometric control of multiple quadrotors transporting a rigid-body load,” in53rd IEEE Conference on Decision and Control. IEEE, 2014, pp. 6141–6148

    2014

  29. [29]

    Solving multi- agent safe optimal control with distributed epigraph form MARL,

    S. Zhang, O. So, M. Black, Z. Serlin, and C. Fan, “Solving multi- agent safe optimal control with distributed epigraph form MARL,” in Proceedings of Robotics: Science and Systems, 2025

    2025

  30. [30]

    Automated aerial suspended cargo delivery through reinforcement learning,

    A. Faust, I. Palunko, P. Cruz, R. Fierro, and L. Tapia, “Automated aerial suspended cargo delivery through reinforcement learning,”Artificial Intelligence, vol. 247, pp. 381–398, 2017

    2017

  31. [31]

    Deep reinforcement learning-based swing-free trajectories planning algorithm for uav with a suspended load,

    R. Li, F. Yang, Y . Xu, W. Yuan, and Q. Lu, “Deep reinforcement learning-based swing-free trajectories planning algorithm for uav with a suspended load,” in2022 China Automation Congress (CAC). IEEE, 2022, pp. 6149–6154

    2022

  32. [32]

    Reinforcement learning based trajectory planning for multi- uav load transportation,

    J. Estevez, J. Manuel Lopez-Guede, J. del Valle-Echavarri, and M. Gra ˜na, “Reinforcement learning based trajectory planning for multi- uav load transportation,”IEEE Access, vol. 12, pp. 144 009–144 016, 2024

    2024

  33. [33]

    Dral: Deep reinforcement adaptive learning for multi-uavs navigation in unknown indoor environment,

    K. Mo, L. Chu, X. Zhang, X. Su, Y . Qian, Y . Ou, and W. Pretorius, “Dral: Deep reinforcement adaptive learning for multi-uavs navigation in unknown indoor environment,” in2024 7th International Conference on Mechatronics and Computer Technology Engineering (MCTE), 2024, pp. 859–863

    2024

  34. [34]

    Cooperative payload transportation by uavs: A model- based deep reinforcement learning (mbdrl) application,

    S. A. Khursheed, “Cooperative payload transportation by uavs: A model- based deep reinforcement learning (mbdrl) application,” Master’s thesis, Virginia Tech, 2024

    2024

  35. [35]

    Multi-uav cooperative transportation using dy- namic control allocation and a reinforcement learning compensator,

    S. Li and D. Zanotto, “Multi-uav cooperative transportation using dy- namic control allocation and a reinforcement learning compensator,” in International Design Engineering Technical Conferences and Computers and Information in Engineering Conference, vol. 85468. American Society of Mechanical Engineers, 2021, p. V009T09A034

    2021

  36. [36]

    Decentralized aerial manipulation of a cable-suspended load using multi-agent reinforcement learning,

    J. Zeng, A. M. Gimenez, E. Vinitsky, J. Alonso-Mora, and S. Sun, “Decentralized aerial manipulation of a cable-suspended load using multi-agent reinforcement learning,” inConference on Robot Learning. PMLR, 2025, pp. 3850–3868

    2025

  37. [37]

    Discrete GCBF proximal policy optimization for multi-agent safe optimal control,

    S. Zhang, O. So, M. Black, and C. Fan, “Discrete GCBF proximal policy optimization for multi-agent safe optimal control,” inThe Thirteenth International Conference on Learning Representations, 2025

    2025

  38. [38]

    GCBF+: A neural graph control barrier function framework for distributed safe multiagent control,

    S. Zhang, O. So, K. Garg, and C. Fan, “GCBF+: A neural graph control barrier function framework for distributed safe multiagent control,”IEEE Transactions on Robotics, vol. 41, pp. 1533–1552, 2025

    2025

  39. [39]

    Control barrier functions: Theory and applications,

    A. D. Ames, S. Coogan, M. Egerstedt, G. Notomista, K. Sreenath, and P. Tabuada, “Control barrier functions: Theory and applications,” in2019 18th European control conference (ECC). Ieee, 2019, pp. 3420–3431

    2019

  40. [40]

    Control barrier functions and input-to-state safety with application to automated vehicles,

    A. Alan, A. J. Taylor, C. R. He, A. D. Ames, and G. Orosz, “Control barrier functions and input-to-state safety with application to automated vehicles,”IEEE Transactions on Control Systems Technology, vol. 31, no. 6, pp. 2744–2759, 2023

    2023

  41. [41]

    Scalable multi-agent reinforcement learning through intelligent information aggregation,

    S. Nayak, K. Choi, W. Ding, S. Dolan, K. Gopalakrishnan, and H. Balakrishnan, “Scalable multi-agent reinforcement learning through intelligent information aggregation,” inInternational conference on machine learning. PMLR, 2023, pp. 25 817–25 833

    2023

  42. [42]

    Multi-agent constrained policy optimisation,

    S. Gu, J. G. Kuba, M. Wen, R. Chen, Z. Wang, Z. Tian, J. Wang, A. Knoll, and Y . Yang, “Multi-agent constrained policy optimisation,” arXiv preprint arXiv:2110.02793, 2021

    Pith/arXiv arXiv 2021

  43. [43]

    Safe multi-agent reinforcement learning for multi-robot control,

    S. Gu, J. G. Kuba, Y . Chen, Y . Du, L. Yang, A. Knoll, and Y . Yang, “Safe multi-agent reinforcement learning for multi-robot control,”Artificial Intelligence, vol. 319, p. 103905, 2023

    2023