Pith. sign in

REVIEW 4 major objections 6 minor 26 references

DeReCo: Decoupling Representation and Coordination Learning for Object-Adaptive Decentralized Multi-Robot Cooperative Transport

T0 review · 4 major / 6 minor · reviewed 2026-08-02 · deepseek-v4-flash

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

Pith's one-line read DeReCo argues that decoupling object representation learning from coordination learning—through a three-stage privileged-training, supervised-reconstruction, and fine-tuning pipeline—lets two mobile manipulators transport objects they have

desk verdict A plausible three-stage recipe for object-adaptive multi-robot transport with solid experimental evidence, but the paper never defines the central object-dependent representation g that Stage 2 reconstructs, leaving the core mechanism untestable as written. read the letter →

arxiv 2603.08111 v2 pith:FSWXWZBG submitted 2026-03-09 cs.RO

Kazuki Shibata , Ryosuke Sota , Shandil Dhiresh Bosch , Yuki Kadokawa , Yoshihisa Tsurumine , Takamitsu Matsubara This is my paper · ORCID
classification cs.RO
keywords multi-robotcooperativetransportdecentralizedexecutionmulti-agentreinforcementlearningobjectgeneralizationrepresentationprivilegedinformationadaptiveencodersim-to-realtransfer
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper claims that the reason robot teams struggle to carry new objects is that today's end-to-end training forces two difficult learning problems—inferring what the object is from partial sensor data, and learning to coordinate with another robot—to be solved simultaneously, so each interferes with the other. Its proposed remedy, DeReCo, runs three stages: first train coordinated behavior while giving the team the object's mass, friction, and shape as privileged information; then train a separate adaptive encoder to reconstruct the object-dependent representation from each robot's own local observations; finally fine-tune the coordination policy using only those reconstructed representations. The reported experiments show higher success rates on three training shapes and six unseen shapes in simulation, and successful zero-shot transfer to two unseen objects on real robots. A reader should care because the recipe, if it holds, is a general way to add object generalization to decentralized manipulation without inventing new learning algorithms.

What carries the argument

The key mechanism is the adaptive encoder: an LSTM trained by supervised regression to map each robot's 27-dimensional local observation—previous action, arm joint angles, gripper states, offset to the goal, relative object pose, and quantized force readings—to the object-dependent representation g_i^t that Stage 1's network computed from privileged object information. That representation is never explicitly defined; the paper treats it as whatever internal code Stage 1's encoder produces. The adaptive encoder is this code's decoder, and it is the piece that lets the final policy act without privileged information at execution time.

What would settle it

Train DeReCo with a deliberately broken Stage-2 target (random vectors, or a constant) and check whether Stage-3 performance collapses; alternatively, measure Stage-2 reconstruction error on each unseen object and correlate it per-object with success rate—if episodes succeed despite poor reconstruction, the reconstructed representation is not what carries the generalization.

Watch

Extended reading notes

Core claim

The central discovery is the empirical claim that decoupling removes 'bidirectional interference': under partial observability, a robot's inferred representation of the object is noisy, and that noise destabilizes coordination learning; at the same time, the non-stationary multi-agent learning dynamics corrupt the representation learner. DeReCo's three-stage training gives each stage a well-posed objective. Stage 1 establishes coordination under reliable object knowledge; Stage 2 solves a supervised regression problem from local observations to the representation Stage 1 used; Stage 3 fine-tunes the policy with the learned encoder and without privileged information. On the paper's measuremen

Load-bearing premise

The whole pipeline rests on the assumption that the object-dependent representation g_i^t used as Stage-2 supervision is well-defined, inferable from a robot's 27-dimensional local observation by a frozen LSTM, and sufficient for coordination without privileged information; Section IV-C never defines what g_i^t actually encodes.

Editorial extensions

If this is right

  • If the decoupling claim holds, end-to-end joint training with domain randomization is not the only route to object generalization in multi-robot transport; a staged privileged-then-reconstruction pipeline can be more sample-efficient.
  • At execution, each robot needs only its own local observations—no shared state, no communication, no object model or mass/friction measurement.
  • The generalization result suggests that within the trained ranges of mass and friction, a single policy can transport unseen shapes with useful success rates, not just the exact training objects.
  • Success on two unseen objects in hardware without any real-world fine-tuning indicates the representation learned in simulation carries over to real sensors.
  • The failure analysis suggests that remaining errors are mostly transport-precision failures rather than grasp failures, pointing at coordination accuracy as the next bottleneck.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A broader recipe follows if the paper is right: for any MARL task with a latent environment factor, first learn with oracle access to that factor, distill the factor into a local encoder, then fine-tune the policy; the paper's own design is one instantiation of that recipe rather than a special-purpose transport trick.
  • The paper leaves the supervision target g_i^t undefined, so a fair test of the framework would be to vary what Stage 2 reconstructs (object identity vs. low-dimensional physical parameters vs. the raw Stage-1 code) and watch success rates; this would show which notion of 'object-dependent representation' actually carries the behavior.
  • A testable extension is to apply the same three-stage scheme to teams larger than two or to objects that change properties mid-episode; the paper's discussion notes that its current MAPPO-based policy input depends on the number of robots.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. DeReCo proposes a three-stage MARL framework for decentralized two-robot cooperative transport. Stage 1 trains MAPPO-style actors and a centralized critic with privileged object information (mass, friction, shape); Stage 2 trains an LSTM-based adaptive encoder to reconstruct an 'object-dependent representation' g_i^t from each robot's 27-dimensional local observation; Stage 3 fine-tunes the actor with the frozen adaptive encoder while the critic retains privileged information, then deploys actors from local observations only. The paper evaluates the method in Isaac Sim on three training objects and six unseen objects, comparing against five MAPPO baselines, and reports real-robot trials on two unseen objects. The central claim is that this decoupling of representation learning from coordination learning mitigates bidirectional interference and improves sample efficiency and object generalization.

Significance. If the claimed results hold, DeReCo is a practically useful design pattern for object-adaptive decentralized manipulation: it combines privileged-info pretraining, supervised representation distillation, and fine-tuning in a way that is straightforward to implement with standard MAPPO. The paper has several concrete strengths: it reports success rates over 1,000 trials per policy for nine simulated objects, provides a failure-mode analysis, and includes real-robot transfer experiments with two unseen objects. The staged training is not circular: Stage 2 is supervised by targets from the Stage-1 encoder, and success rates are measured independently from the training objective. However, the central mechanism is underspecified: the target quantity g_i^t is never defined, and the quantitative evidence for sample efficiency and for 'outperforming' baselines is incomplete. These issues are load-bearing because they prevent reproduction and because they leave the causal attribution to 'decoupling' unsupported.

major comments (4)
  1. [Section IV-C, Fig. 2] The supervised target g_i^t is never defined. The text says only that g_i^t is 'computed during rollouts using the encoder in Stage 1,' but Stage 1's architecture (Section V-A3) is described only as FC layers with 128 units; no layer or output is identified as g_i^t, and no dimensionality, normalization, or semantic content is stated. This matters because Stage 2's MSE objective L(g_i^t, g_hat_i^t) and Stage 3's use of the trained encoder are the core mechanism by which privileged information is supposed to be removed. Without a definition, a reader cannot determine whether g_i^t is recoverable from the 27-dimensional local observation or whether it encodes global/privileged quantities (e.g., exact friction, mass, or object identity) that are not identifiable from local force/pose signals. The authors should specify exactly which encoder output is used as g_i^t, its information content,
  2. [Table I, Section V-C] All success rates in Table I are point estimates without error bars, confidence intervals, or significance tests, despite being averaged over five independently trained policies and 1,000 trials each. The claim that DeReCo 'outperforms' baselines is not fully supported by the reported numbers: on the seen Board, DeReCo (0.94) is slightly below MAPPO w/o AE (0.95), and on seen-object average DeReCo and MAPPO w PI are tied at 0.91. Several unseen-object differences are also within plausible seed noise. The authors should report standard deviations, bootstrap confidence intervals, or paired significance tests (e.g., across the five seeds) before claiming consistent superiority.
  3. [Section V-B2, Fig. 4] The sample-efficiency claim rests on a single reward curve without variance or a quantitative metric. Fig. 4 shows only a mean curve for each method over 50,000 training steps. More importantly, the DeReCo curve is for Stage 3 only; Stages 1 and 2 consume additional environment interactions before Stage 3 begins. Comparing Stage 3's curve to end-to-end baselines that also train for 50,000 steps does not compare total sample budgets. The authors should state the number of environment steps used in Stages 1 and 2, and report a metric such as area under the reward curve or steps to a success-rate threshold computed over the full training pipeline.
  4. [Section IV-D] Stage 3 is described as 'progressively' removing privileged information, but no progressive schedule, curriculum, or annealing is defined. From the text, it appears that the actor is simply reinitialized from Stage 1 weights and retrained with the frozen adaptive encoder, while the critic continues to use privileged information. If there is no gradual removal, the word 'progressive' is misleading; if there is a schedule, it should be specified. This matters because the claimed decoupling mechanism relies on the manner in which privileged information is removed.
minor comments (6)
  1. [Section V-A2] The baseline 'MAPPO w/o AE' is used as the strongest comparison in the real-robot experiments, but its architecture is not fully specified relative to DeReCo's actor. Please clarify whether it uses the same local-observation encoder and MLP layers, without the adaptive encoder, so that the comparison isolates the proposed mechanism.
  2. [Section V-A] PPO/MAPPO hyperparameters (learning rate, discount factor, GAE lambda, clip ratio, entropy coefficient, minibatch size, number of epochs) and the training budget for each stage are not reported. These are needed for reproducibility and for interpreting the sample-efficiency comparison.
  3. [Section V-A1] The observation is said to be in R^27, but the listed components sum to 6+3+2+3+7+6 = 27. Please make the decomposition explicit in the text, as some readers may mis-count the previous action or the force channels.
  4. [Table II] In Table II, MAPPO w/o AE on Board reports 0/5 success but a final error of 0.18 +/- 0.06 m; on Frame it reports 0.34 +/- 0.00 m. Please clarify how the error is computed for failed trials, and consider reporting per-trial values or a full success/failure breakdown.
  5. [General] The demonstration video is referenced as 'link' with no actual URL. If the video is intended to be part of the submission, provide a working link or remove the reference.
  6. [Section IV-C] The notation for the mean squared error L(g_i^t, g_hat_i^t) is introduced but the loss is not written as an equation. A short equation, including whether it is averaged over time and over the batch of collected rollouts, would improve precision.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: DeReCo's three-stage distillation is not forced by construction; only minor contextual self-citations and an underspecified Stage-2 target.

full rationale

DeReCo's derivation chain is empirical and self-contained against external benchmarks. Stage 2 trains an adaptive encoder eψ to reconstruct g_i^t, an internal representation 'computed during rollouts using the encoder in Stage 1' (Sec. IV-C); Stage 3 fine-tunes the policy with that frozen encoder. This is teacher-student distillation, not a definitional identity: the downstream success rates (Tables I–II) are independent measurements on held-out objects and real robots. No equation in the paper reduces a predicted quantity to a fitted parameter by construction, and no uniqueness theorem or prior-work ansatz is invoked to force the design. The self-citations—[1] for ternary force observation/reward design and [6] as future scalability—are contextual and do not carry the central claim. The main weakness is that g_i^t is never formally defined (which layer, dimension, or information content), making Stage 2 hard to reproduce; that is a soundness gap, not circularity.

Assumptions & free parameters 3 free parameters · 5 assumptions · 1 invented entities

The central claim rests on the existence and sufficiency of an undefined latent representation g, on simulator fidelity for sim-to-real transfer, and on the untested assumption that a three-stage training schedule can distill coordination knowledge. No free parameters are reported beyond unspecified hyperparameters, and no formal verification is provided.

free parameters (3)
  • PPO/MAPPO hyperparameters (learning rate, discount factor, GAE lambda, clip ratio, entropy coefficient)
    Not reported in the paper; chosen by hand and likely tuned per task. These affect training stability and sample efficiency, which are central claims.
  • Network hidden sizes (128-unit FC layers and 128-unit LSTM)
    Selected by hand, described in Section V-A3; impact representation capacity and generalization.
  • Training budget (50,000 steps, 512 parallel environments)
    Chosen by hand for the 512-parallel setting, Section V-A and V-B; affects the sample-efficiency comparison.
assumptions (5)
  • domain assumption The Isaac Sim physics simulator accurately models the contact dynamics and object properties needed for sim-to-real transfer.
    The paper's empirical claims rely on the fidelity of simulation (Section V-A); no domain gap analysis is provided.
  • domain assumption The local observation o_i (R^27) contains sufficient information to infer the object-dependent representation and coordinate transport.
    Stated in Section III-A; if the observation space omits necessary information (e.g., other robot's action or contact forces), the decentralized task may be unlearnable.
  • domain assumption The reward design from [1] is a suitable shaping reward for cooperative transport and does not bias the comparison.
    Section V-A1: 'We use the same reward design as in [1]' with no sensitivity analysis.
  • standard math MAPPO with the clipped surrogate objective (Eq. 1) converges to a stable coordination policy; standard PPO convergence assumptions hold in the multi-agent setting.
    MAPPO is presented as the base algorithm (Section III-C); convergence is assumed without proof.
  • ad hoc to paper The three-stage training schedule—privileged MARL, supervised reconstruction, fine-tuning—can distill coordination knowledge without catastrophic forgetting.
    The core method of the paper; no theoretical justification is given. Introduced ad hoc in Section IV-A.
invented entities (1)
  • Object-dependent representation g_i^t
    purpose: A latent vector computed by the Stage-1 encoder from privileged information, used as the supervision target for the Stage-2 adaptive encoder and as the input for the decentralized policy in Stage 3.
    Despite being the central quantity of the method, g is never defined mathematically or semantically. The paper states it is 'computed during rollouts using the encoder in Stage 1' (Section IV-C) but does not specify its dimension, content, or why it is inferable from local observations. No external falsifiable handle is provided.

how reviews work

0 comments
Cite this review

Pith. "Pith review of DeReCo: Decoupling Representation and Coordination Learning for Object-Adaptive Decentralized Multi-Robot Cooperative Transport." pith.science (2026). https://pith.science/paper/FSWXWZBG

@misc{pith2026260308111,
  author       = {Pith},
  title        = {Pith review of: DeReCo: Decoupling Representation and Coordination Learning for Object-Adaptive Decentralized Multi-Robot Cooperative Transport},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FSWXWZBG}},
  note         = {Machine review of arXiv:2603.08111}
}
read the original abstract

Generalizing decentralized multi-robot cooperative transport across objects with diverse shapes and physical properties remains a fundamental challenge. Under decentralized execution, two key challenges arise: object-dependent representation learning under partial observability and coordination learning in multi-agent reinforcement learning (MARL) under non-stationarity. A typical approach jointly optimizes object-dependent representations and coordinated policies in an end-to-end manner while randomizing object shapes and physical properties during training. However, this joint optimization tightly couples representation and coordination learning, introducing bidirectional interference: inaccurate representations under partial observability destabilize coordination learning, while non-stationarity in MARL further degrades representation learning, resulting in sample-inefficient training. To address this structural coupling, we propose DeReCo, a novel MARL framework that decouples representation and coordination learning for object-adaptive multi-robot cooperative transport, improving sample efficiency and generalization across objects and transport scenarios. DeReCo adopts a three-stage training strategy: (1) centralized coordination learning with privileged object information, (2) reconstruction of object-dependent representations from local observations, and (3) progressive removal of privileged information for decentralized execution. This decoupling mitigates interference between representation and coordination learning and enables stable and sample-efficient training. Experimental results show that DeReCo outperforms baselines in simulation on three training objects, generalizes to six unseen objects with varying masses and friction coefficients, and achieves superior performance on two unseen objects in real-robot experiments.

Figures

Figures reproduced from arXiv: 2603.08111 by the authors.

Figure 1
Figure 1. Multi-robot cooperative transport of diverse objects [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Overview of the DeReCo for object-adaptive decentralized multi-robot cooperative transport [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Objects used in the simulation. All units are meters. [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Comparisons of the tracking reward w/o AE + LSTM, exhibit lower rewards, suggesting that simply introducing recurrence into end-to-end training does not resolve the structural coupling between representation learning and coordination learning. Overall, DeReCo can impro…
Figure 5
Figure 5. Figure 5: Failure analysis on unseen objects for four representative methods [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Objects used in real experiments. All units are meters. [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: Comparisons of real-robot demonstrations. [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

26 extracted references · 2 linked inside Pith

  1. [22]

    Learning whole-body manipulation for quadrupedal robot,

    S. Jeon, M. Jung, S. Choi, B. Kim, and J. Hwangbo, “Learning whole-body manipulation for quadrupedal robot,”IEEE Robotics and Automation Letters, vol. 9, no. 1, pp. 699–706, 2024

    2024

  2. [1]

    Cooperative grasping and transportation using multi-agent reinforcement learning with ternary force representation,

    I.-S. Bernard-Tiong, Y . Tsurumine, R. Sota, K. Shibata, and T. Mat- subara, “Cooperative grasping and transportation using multi-agent reinforcement learning with ternary force representation,” in2025 IEEE/SICE International Symposium on System Integration (SII), 2025, pp. 973–978

    2025

  3. [2]

    Learning multi-agent loco-manipulation for long-horizon quadrupedal pushing,

    Y . Feng, C. Hong, Y . Niu, S. Liu, Y . Yang, and D. Zhao, “Learning multi-agent loco-manipulation for long-horizon quadrupedal pushing,” in2025 IEEE International Conference on Robotics and Automation (ICRA), 2025, pp. 14 441–14 448

    2025

  4. [3]

    Learning decentralized multi-biped control for payload transport,

    B. Pandit, A. Gupta, M. S. Gadde, A. Johnson, A. K. Shrestha, H. Duan, J. Dao, and A. Fern, “Learning decentralized multi-biped control for payload transport,” inProceedings of The 8th Conference on Robot Learning, vol. 270. PMLR, 2025, pp. 1021–1034

    2025

  5. [4]

    Deep reinforcement learning for coordinated payload transport in biped-wheeled robots,

    D. K. Mehta, A. Joglekar, and V . Krovi, “Deep reinforcement learning for coordinated payload transport in biped-wheeled robots,” in2025 IEEE International Conference on Robotics and Automation (ICRA), 2025, pp. 14 992–14 998. 1 1 0.5 0 0.5 0 Board Grasp ✘ FailureInitial state Frame Initial state Grasp Success✘ Failure ★ Goal 〇 Start ー Trajectory 0 -0.5 0...

    2025

  6. [5]

    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,” inProceedings of The 9th Con- ference on Robot Learning, vol. 305. PMLR, 2025, pp. 3850–3868

    2025

  7. [6]

    Deep reinforcement learning of event-triggered communication and consensus-based control for distributed cooperative transport,

    K. Shibata, T. Jimbo, and T. Matsubara, “Deep reinforcement learning of event-triggered communication and consensus-based control for distributed cooperative transport,”Robotics and Autonomous Systems, vol. 159, p. 104307, 2023

    2023

  8. [7]

    De- centralized navigation of a cable-towed load using quadrupedal robot team via MARL,

    W.-T. Chen, M. Nguyen, Z. Li, G. N. Sue, and K. Sreenath, “De- centralized navigation of a cable-towed load using quadrupedal robot team via MARL,”arXiv preprint arXiv:2503.18221, 2025

    arXiv 2025

Show all 26 references
  1. [8]

    Multi-agent actor-critic for mixed cooperative-competitive environments,

    R. Lowe, Y . WU, A. Tamar, J. Harb, O. Pieter Abbeel, and I. Mor- datch, “Multi-agent actor-critic for mixed cooperative-competitive environments,” inAdvances in Neural Information Processing Systems, vol. 30, 2017

    2017

  2. [9]

    Multi-robot object transport motion planning with a deformable sheet,

    J. Hu, W. Liu, H. Zhang, J. Yi, and Z. Xiong, “Multi-robot object transport motion planning with a deformable sheet,”IEEE Robotics and Automation Letters, vol. 7, no. 4, pp. 9350–9357, 2022

    2022

  3. [10]

    Coopera- tive multi-robot object transportation system based on hierarchical quadratic programming,

    D. Koung, O. Kermorgant, I. Fantoni, and L. Belouaer, “Coopera- tive multi-robot object transportation system based on hierarchical quadratic programming,”IEEE Robotics and Automation Letters, vol. 6, no. 4, pp. 6466–6472, 2021

    2021

  4. [11]

    Payload-aware trajectory optimisa- tion for non-holonomic mobile multi-robot manipulation with tip-over avoidance,

    F. Kennel-Maushart and S. Coros, “Payload-aware trajectory optimisa- tion for non-holonomic mobile multi-robot manipulation with tip-over avoidance,”IEEE Robotics and Automation Letters, vol. 9, no. 9, pp. 7669–7676, 2024

    2024

  5. [12]

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

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

    2023

  6. [13]

    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

  7. [14]

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

    R. T. Fawcett, L. Amanzadeh, J. Kim, A. D. Ames, and K. A. Hamed, “Distributed data-driven predictive control for multi-agent collabora- tive legged locomotion,” in2023 IEEE International Conference on Robotics and Automation (ICRA), 2023, pp. 9924–9930

    2023

  8. [15]

    Distributed NMPC for cooperative aerial manipulation of cable-suspended loads,

    N. De Carli, R. Belletti, E. Buzzurro, A. Testa, G. Notarstefano, and M. Tognon, “Distributed NMPC for cooperative aerial manipulation of cable-suspended loads,”IEEE Robotics and Automation Letters, vol. 10, no. 10, pp. 10 546–10 553, 2025

    2025

  9. [16]

    Decentralized adaptive control for collaborative manipulation of rigid bodies,

    P. Culbertson, J.-J. Slotine, and M. Schwager, “Decentralized adaptive control for collaborative manipulation of rigid bodies,”IEEE Trans- actions on Robotics, vol. 37, no. 6, pp. 1906–1920, 2021

    1906

  10. [17]

    Decentralized ability-aware adaptive control for multi-robot collaborative manipulation,

    L. Yan, T. Stouraitis, and S. Vijayakumar, “Decentralized ability-aware adaptive control for multi-robot collaborative manipulation,”IEEE Robotics and Automation Letters, vol. 6, no. 2, pp. 2311–2318, 2021

    2021

  11. [18]

    Hierarchical adaptive control for collaborative manipulation of a rigid object by quadrupedal robots,

    M. Sombolestan and Q. Nguyen, “Hierarchical adaptive control for collaborative manipulation of a rigid object by quadrupedal robots,” in2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2023, pp. 2752–2759

    2023

  12. [19]

    Sim-to- real transfer of robotic control with dynamics randomization,

    X. B. Peng, M. Andrychowicz, W. Zaremba, and P. Abbeel, “Sim-to- real transfer of robotic control with dynamics randomization,” in2018 IEEE International Conference on Robotics and Automation (ICRA), 2018, pp. 3803–3810

    2018

  13. [20]

    COCOI: Contact-aware online context inference for generalizable non-planar pushing,

    Z. Xu, W. Yu, A. Herzog, W. Lu, C. Fu, M. Tomizuka, Y . Bai, C. K. Liu, and D. Ho, “COCOI: Contact-aware online context inference for generalizable non-planar pushing,” in2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2021, pp. 176– 182

    2021

  14. [21]

    Dynamic object goal pushing with mobile manipulators through model-free constrained reinforcement learning,

    I. Dadiotis, M. Mittal, N. Tsagarakis, and M. Hutter, “Dynamic object goal pushing with mobile manipulators through model-free constrained reinforcement learning,” in2025 IEEE International Conference on Robotics and Automation (ICRA), 2025, pp. 13 363–13 369

    2025

  15. [23]

    F. A. Oliehoek and C. Amato,A Concise Introduction to Decentralized POMDPs, 1st ed. Springer Publishing Company, Incorporated, 2016

    2016

  16. [24]

    The surprising effectiveness of PPO in cooperative multi- agent games,

    C. Yu, A. Velu, E. Vinitsky, J. Gao, Y . Wang, A. Bayen, and Y . Wu, “The surprising effectiveness of PPO in cooperative multi- agent games,” inAdvances in Neural Information Processing Systems, vol. 35, 2022, pp. 24 611–24 624

    2022

  17. [25]

    Proximal policy optimization algorithms,

    J. Schulman, F. Wolski, P. Dhariwal, A. Radford, and O. Klimov, “Proximal policy optimization algorithms,”arXiv preprint arXiv:1707.06347, 2017

    2017 arXiv

  18. [26]

    Domain randomization via entropy maximization,

    G. Tiboni, P. Klink, J. Peters, T. Tommasi, C. D’Eramo, and G. Chal- vatzaki, “Domain randomization via entropy maximization,” inThe Twelfth International Conference on Learning Representations, 2024

    2024

Pith tools

Reviewed August 2, 2026 · model on record in the stance chip above.