Pith. sign in

REVIEW 3 major objections 3 minor 6 cited by

CycleVLA claims that wrapping a vision-language-action model with progress-aware failure prediction, subtask backtracking, and consensus retry lifts success rates on long-horizon manipulation tasks.

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 ↗

CycleVLA adds progress-triggered VLM failure checks, subtask backtracking, and MBR consensus decoding to VLAs, raising LIBERO average success from 89.3% to 95.3% and claiming 91% real-robot success.

T0 review reviewed 2026-08-03 challenge →

load-bearing objection Solid simulation wrapper with a genuine MBR-for-VLA idea, but the abstract claims real-robot results the paper itself says are not yet done. the 3 major comments →

arxiv 2601.02295 v2 pith:JXNS6OPV submitted 2026-01-05 cs.RO

CycleVLA: Proactive Self-Correcting Vision-Language-Action Models via Subtask Backtracking and Minimum Bayes Risk Decoding

Chenyang Ma , Kai Lu , Guangyu Yang , Jiuming Liu , Shitong Xu , Bill Byrne , Ioannis Havoutis , Niki Trigoni
show 1 more author
Andrew Markham
This is my paper
classification cs.RO
keywords vision-language-action modelsproactive self-correctionfailure predictionsubtask backtrackingminimum Bayes risk decodingtest-time scalingrobot manipulationlong-horizon tasks
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.

The reading

The paper argues that robot failures can be caught before they fully manifest by teaching a vision-language-action model to estimate its own subtask progress, having a vision-language model judge whether the current subtask will fail, and if so rewinding to an earlier subtask and retrying with a consensus-selected action. It claims this wrapping improves success rates for both fully trained and under-trained policies, most strongly on long-horizon tasks, and that Minimum Bayes Risk decoding is an effective zero-shot test-time scaling strategy. A reader should care because it offers a way to make generalist robot policies more reliable without retraining the underlying policy.

Core claim

The central claim is that equipping a VLA with explicit subtask progress and stop signals, then querying a VLM at ~90% progress to predict failure and, if predicted, backtracking to the earliest subtask that restores missing preconditions and retrying with MBR-decoded consensus actions, yields large success-rate improvements. The paper shows average success on the LIBERO simulation benchmark rises from 89.3% without correction to 95.3% with correction, with gains across under-trained checkpoints, and reports a 91% average success rate on a real robot on precise and long-horizon manipulation tasks while correcting about 80% of injected failures.

What carries the argument

The central mechanism is a progress-aware VLA: the action space is extended from 7 to 9 dimensions, adding a per-step stop signal and a discretized progress value, trained on a subtask-decomposed dataset. At inference, when predicted progress reaches a threshold (τ_p=0.9), an off-the-shelf VLM is queried to decide whether to transit or backtrack, with backtracking executed by reverse-executing recorded delta actions. After backtracking, MBR decoding samples N=8 stochastic action chunks and selects the medoid of the densest cluster under L2 trajectory distance, a training-free consensus selection that favors high-density regions of the policy output space.

Load-bearing premise

The load-bearing premise is that backtracking can restore the robot to the exact start of a subtask by reversing previously executed actions; if the world does not reverse cleanly (contacts, friction, object shifts), the retried state differs from the recorded one and the VLM's plan may no longer apply.

What would settle it

Run CycleVLA on a task where an object is knocked over or a grasped item is dropped and displaced during the failed attempt, and record whether success after backtracking drops to the no-correction baseline; if it does, reverse execution is not actually restoring preconditions.

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

If this is right

  • If correct, any stochastic VLA can be wrapped with progress prediction, VLM-gated backtracking, and MBR retry to gain several points of success without retraining the policy.
  • Test-time scaling via MBR becomes a zero-shot alternative to learned verifiers or reward models for VLA action selection.
  • Under-trained policies benefit more, suggesting that inference-time consensus can partially compensate for training deficiency.
  • Long-horizon tasks, where errors accumulate across subtasks, stand to gain the most from this self-correction loop.

Where Pith is reading between the lines

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

  • The reversibility assumption means the approach is limited to near-static scenes; in contact-rich or irreversible settings, reverse-executing recorded actions will not restore preconditions, so the reported gains should shrink or vanish.
  • The paper's ablation shows a ~10% drop when terminating on predicted failure, consistent with VLM sycophancy; a calibrated failure predictor would likely recover some of that gap.
  • The density-based MBR selection suggests consensus in action space correlates with success; extending this to visual state consensus or trajectory-level features could improve robustness.
  • One could test whether MBR's benefit transfers to other action representations (e.g., waypoint deltas) and whether using an oracle success label confirms that consensus corresponds to ground-truth success.
Share X Bluesky LinkedIn Reddit HN

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

3 major / 3 minor

Summary. The paper proposes CycleVLA, a system that wraps a pretrained Vision-Language-Action model (VLA) with three components: (a) finetuning that augments the VLA action space with stop and progress signals learned from LLM-decomposed demonstration subtasks; (b) a VLM-based failure predictor/planner that is triggered when predicted progress is high and may command backtracking to an earlier subtask; and (c) Minimum Bayes Risk (MBR) decoding to select among sampled action chunks after backtracking. Experiments are reported on LIBERO simulation, with average success improving from 89.3% to 95.3% for the 500K checkpoint and consistent gains for earlier checkpoints, together with ablations and a runtime analysis. The abstract additionally claims real-robot 91% success, ~80% recovery from injected perturbations, LIBERO-Plus results, and superiority over the π0.5 VLA; these claims do not appear in Section V and are directly contradicted by the Conclusion's statement that real-robot experiments are to be added in the near future.

Significance. If the simulation results are taken at face value, the core idea is practically valuable and timely: a generic VLA can be augmented with progress-triggered VLM failure prediction, subtask backtracking, and MBR consensus selection to improve long-horizon success without training a replacement policy. The paper is also commendably detailed in its appendices, providing exact prompts, hyperparameters, and a human evaluation of the subtask-decomposition pipeline. However, the central empirical contribution as advertised—real-robot 91% success, ~80% stress-test recovery, LIBERO-Plus evaluation, and superiority over π0.5—is not verifiable from the submitted text. The verifiable contribution is a LIBERO-only simulation study, and even that rests on an unexamined reversibility assumption for the backtracking mechanism.

major comments (3)
  1. [Abstract vs. Section V and Conclusion] The abstract advertises (i) a 91% average success rate on a real robot, (ii) correction of ~80% of injected failures in stress tests, (iii) evaluation on LIBERO and LIBERO-Plus simulation benchmarks, and (iv) superiority over the state-of-the-art VLA π0.5. None of these are present in Section V: Tables I–VI report only LIBERO; Table I contains no π0.5 row; no stress-test protocol or perturbation results appear anywhere; and the Conclusion explicitly states 'Experiments on real robot hardware to be added in the near future.' The headline empirical claims are therefore unsupported and internally inconsistent with the manuscript's own text, and the central claim as stated cannot be evaluated.
  2. [Section IV-B, Algorithm 1 line 17, and Conclusion] The backtracking mechanism restores the robot to the start of subtask g_j by 'reverse-executing recorded delta actions.' This assumes reversible state transitions. In contact-rich manipulation, objects may be displaced, grasps may drop objects, or the robot state may drift during the failed attempt, so the retry can begin from a state different from the recorded start. The Conclusion acknowledges this limitation ('assumes reversible state transitions, which may not hold in highly dynamic or irreversible environments'). Because every reported recovery gain depends on this assumption, the paper needs evidence of state-restoration fidelity (e.g., measured state deviation after reverse execution, or success rates conditioned on the actual post-backtracking state) before the gains can be attributed to the proposed correction mechanism.
  3. [Section V-C, Eq. (5)] The MBR evaluation metric P_succ estimates success probability by averaging per-chunk success labels z_e,t 'inherited from the recorded rollout.' But if MBR selects a different hypothesis at an early decision step, the subsequent observations, action outcomes, and success labels would not be those recorded in that rollout. Thus Eq. (5) is not a valid closed-loop success estimate; it is an open-loop selection metric that can be optimistic. The end-to-end gains in Table II combine MBR with backtracking, so the isolated MBR contribution in Table III needs a closed-loop evaluation (execute the selected chunk and continue) or an explicit label as an oracle/upper-bound.
minor comments (3)
  1. [Section V-E] Typo: 'computationn' should be 'computation.'
  2. [Table VI caption] The abbreviation LAO is used in the table caption but not spelled out in the main text; please define at first use.
  3. [Introduction, reference [1]] The term 'proactive self-correction' is cited to a reinforcement-learning intervention paper; the term is not standard in that reference. Consider citing a more directly relevant work or defining the term operationally.

Circularity Check

0 steps flagged

No significant circularity; the correction pipeline is evaluated against an external benchmark and does not reduce to fitted labels or self-citations.

full rationale

The derivation chain is self-contained and externally anchored. The stop and progress signals are defined from a subtask-decomposed demonstration dataset (Sec. IV-A) and predicted by a finetuned VLA; the failure predictor is an off-the-shelf VLM (GPT-5.2) that outputs transit/backtrack from visual evidence, not from the success labels being predicted. MBR (Eqs. 1-4) selects consensus action chunks using pairwise L2 distance over sampled trajectories; it never uses task outcomes, and its contribution is validated by an ablation that removes it (Table VI: 95.3 -> 92.5). No fitted parameter is renamed as a prediction, and no equation is defined in terms of the target success rate. The self-citations present (e.g., refs [26], [38], [44], [79]-[81]) appear in related-work, runtime, or unrelated technical contexts and are not load-bearing for the central claims. The abstract advertises real-robot results, LIBERO-Plus, and a pi0.5 comparison that are absent from Section V and contradicted by the Conclusion ("Experiments on real robot hardware to be added in the near future."); this is a verifiability and correctness concern, not a circularity finding. The paper's declared limitation that backtracking "assumes reversible state transitions" affects validity in dynamic environments but does not make any step circular.

Axiom & Free-Parameter Ledger

8 free parameters · 6 axioms · 1 invented entities

The pipeline introduces several hand-chosen hyperparameters (τp, R, N, r, distance metric, oversampling factor, chunk size) and depends on domain assumptions about subtask boundary detection, VLM failure prediction, reversibility of backtracking, and the clustering of successful actions. None of these are derived from first principles; the central claim rests on empirically chosen components.

free parameters (8)
  • Progress threshold τp = 0.9
    Triggers the VLM failure check at 90% of the (time-based) progress signal; chosen by hand with no ablation on the threshold.
  • Max retries R = 3
    Limits backtracking attempts per subtask in Algorithm 1; chosen by hand.
  • MBR sample count N = 8
    Number of sampled action hypotheses for MBR; Table III shows gains plateau beyond 16, but N=8 is the main setting.
  • r-NN radius r = 2 for N=8
    r = max(2, min(4, floor(sqrt(N)))) in the density-based MBR variant; selected adaptively but effectively a hand-chosen schedule.
  • Distance metric d = L2
    Table IV reports L2 as the best among L1, L2, L∞, cosine, and correlation; chosen post-hoc based on benchmark performance.
  • Per-trajectory translation threshold τ_trans = 0.02 nominal, grid-searched ±0.01
    Appendix B.1 optimizes τ_trans per demonstration to minimize overlaps and spurious stops using score = 1.0*N_overlaps + 2.5*N_stops.
  • Action chunk size H = 8
    Open-loop execution chunk length used in training and inference; chosen without sensitivity analysis.
  • Last-action oversampling factor = 8
    Oversamples the last action step of each subtask during finetuning to emphasize stop-signal prediction.
axioms (6)
  • domain assumption Subtask boundaries are recoverable from gripper-state transition segments plus LLM timestamp inference.
    Section IV-A: the entire finetuning dataset relies on aligning LLM-proposed subtasks to trajectory timestamps using gripper-state segments; if this alignment is wrong, stop/progress supervision is wrong.
  • domain assumption An off-the-shelf VLM can forecast whether a subtask will fail from two camera views at ~90% progress.
    Section IV-B and Appendix F: GPT-5.2 is treated as a reliable zero-shot failure predictor/planner; the only robustness measure is one ablation replacing it with LLaMA-3.2-11B.
  • domain assumption Reverse-executing recorded delta actions restores the exact subtask start state.
    Section IV-B, Algorithm 1 line 17: backtracking restores state by reverse-executing recorded actions; this assumes reversible dynamics, which the Conclusion explicitly flags as a limitation.
  • domain assumption Successful action chunks cluster in high-density regions of the VLA's action output space, so MBR consensus selects successful executions.
    Section IV-B: MBR is motivated by this clustering assumption, citing prior work [32],[33]; the paper does not verify the assumption for OpenVLA's diffusion expert beyond its own benchmarks.
  • domain assumption Repeated stochastic decoding with different random seeds provides diverse and useful action hypotheses from the same observation.
    Section III and V-A: MBR requires sampling from a stochastic policy by varying diffusion noise seeds; diversity and coverage are assumed rather than measured.
  • ad hoc to paper Normalized timestep within a subtask is a valid proxy for physical progress.
    Section IV-A defines p_t as 'normalized timestep within each subtask' but presents it as progress; this conflates elapsed time with completion and is a modeling choice specific to this paper.
invented entities (1)
  • Extended action dimensions: stop signal s_t and progress signal p_t no independent evidence
    purpose: Allow the VLA to signal subtask termination and proximity-to-completion, triggering VLM checks at test time.
    These are new output dimensions introduced in Section IV-A. They are validated only on the paper's own LIBERO experiments and claimed real-robot results, not against external measurements or independent benchmarks.

reviewed 2026-08-03 · how reviews work

0 comments
Cite this review

Pith. "Pith review of CycleVLA: Proactive Self-Correcting Vision-Language-Action Models via Subtask Backtracking and Minimum Bayes Risk Decoding." pith.science (2026). https://pith.science/paper/JXNS6OPV

@misc{pith2026260102295,
  author       = {Pith},
  title        = {Pith review of: CycleVLA: Proactive Self-Correcting Vision-Language-Action Models via Subtask Backtracking and Minimum Bayes Risk Decoding},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JXNS6OPV}},
  note         = {Machine review of arXiv:2601.02295}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Current work on robot failure detection and correction typically operates in a post hoc manner, analyzing errors and applying corrections only after failures occur. This work introduces CycleVLA, a system that equips Vision-Language-Action models (VLAs) with proactive self-correction, the capability to anticipate incipient failures and recover before they fully manifest during execution. CycleVLA achieves this by integrating a progress-aware VLA that flags critical subtask transition points where failures most frequently occur, a VLM-based failure predictor and planner that triggers subtask backtracking upon predicted failure, and a test-time scaling strategy based on Minimum Bayes Risk (MBR) decoding to improve retry success after backtracking. Extensive experiments on the LIBERO and LIBERO-Plus simulation benchmarks show that CycleVLA surpasses the state-of-the-art VLA {\pi}0.5, improves success rates by correcting execution failures across VLAs of varying capability, from under-trained ones to fully converged policies, and that MBR serves as an effective zero-shot test-time scaling strategy for VLAs. On a real robot, CycleVLA reaches a 91% average success rate on one precise and two long-horizon manipulation tasks. We further conduct stress tests with multiple manually injected perturbations (e.g., swapping in a distractor at the expected location while relocating the true target object mid-execution), where CycleVLA corrects ~80% of injected failures and maintains success rates comparable to unperturbed execution. Project Page: https://dannymcy.github.io/cyclevla/

Figures

Figures reproduced from arXiv: 2601.02295 by Andrew Markham, Bill Byrne, Chenyang Ma, Guangyu Yang, Ioannis Havoutis, Jiuming Liu, Kai Lu, Niki Trigoni, Shitong Xu.

Figure 1
Figure 1. Figure 1: Vision–Language–Action models with proactive self-correction capabilities. We introduce CycleVLA, which enables VLAs to anticipate incipient failures and recover before execution collapses. CycleVLA first augments a VLA to estimate subtask-level progress and flag critical subtask transition points, where failures most frequently occur. Then, at these points during inference, a VLM is queried to predict whe… view at source ↗
Figure 2
Figure 2. Figure 2: Pipeline for constructing the subtask-decomposed dataset. Following LLM subtask decomposition and extraction of movement primitives and gripper state segments, subtasks are directly aligned to gripper state segment timestamps when their counts match; otherwise, an LLM infers subtask boundaries from the movement primitive sequence. Please see Appendices B and E for more details. B. Test-Time Scaling via Sub… view at source ↗
Figure 3
Figure 3. Figure 3: CycleVLA. (a) A finetuning pipeline that equips a VLA with subtask-level stop and progress prediction via extended action expert dimension and augmented subtask-decomposed training data. (b) At inference, predicted progress triggers a VLM-based failure predictor and planner, which decides whether to transit to the next subtask or backtrack, and selects the subtask to backtrack to. (c) After backtracking, t… view at source ↗
Figure 4
Figure 4. Figure 4: Qualitative examples of CycleVLA. CycleVLA performs multiple cycles of failure prediction, backtracking, and retry within a single long-horizon task, correcting errors across subtasks and achieving successful completion. More examples can be found in Appendix D. otherwise specified. We use GPT-5.2 [67] (temperature 1.0) as the VLM-based failure predictor and planner, queried when subtask progress reaches τ… view at source ↗
Figure 5
Figure 5. Figure 5: Qualitative examples of subtask-decomposed dataset [PITH_FULL_IMAGE:figures/full_fig_p012_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Additional examples of CycleVLA [PITH_FULL_IMAGE:figures/full_fig_p013_6.png] view at source ↗

discussion (0)

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

Forward citations

Cited by 6 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. ProbeAct: Probe-Guided Training-Free Failure Recovery in Vision-Language-Action Models

    cs.RO 2026-06 unverdicted novelty 7.0

    PROBEACT is a plug-and-play intervention framework that combines hidden-state probing, kinematic failure detection, and CBF-based correction to boost success rates of pre-trained VLA models on the LIBERO-plus benchmar...

  2. RoboBRIDGE: A Modular Framework for Bridging Policies to Robust Real-World Robotic Agents

    cs.RO 2026-07 conditional novelty 6.0

    Wrapping VLA robot policies in a module stack with monitoring, replanning, asynchronous perception, and primitive-specific LoRA adapters improves benchmark success rates, though the headline comparisons mix in task fi...

  3. CAIRN: Cross-Room 3D Scene Understanding with Topology-Aware Large Multimodal Models

    cs.CV 2026-07 conditional novelty 6.0

    A topology-aware 3D-LLM with hierarchical masked attention and geometric bias outperforms prior 3D-LLMs on a new multi-room scene understanding benchmark built from HM3D.

  4. CAIRN: Cross-Room 3D Scene Understanding with Topology-Aware Large Multimodal Models

    cs.CV 2026-07 conditional novelty 6.0

    Topology-aware attention over hierarchical scene graphs lets a 3D-LLM ground, caption, and answer questions across multi-room homes, with large gains on a new HM3D benchmark.

  5. FurnitureVLA: Learning Long-Horizon Bimanual Furniture Assembly with Vision-Language-Action Model

    cs.RO 2026-07 unverdicted novelty 6.0

    Progress-enhanced VLA model raises simulated bimanual furniture assembly success from 48% to 80% across three furniture types and shows 16% drop on real Kinova robot.

  6. VLA-FAIL: Efficient Task Failure Detection for Finetuned Vision-Language-Action Models

    cs.LG 2026-06 unverdicted novelty 6.0

    VLA-FAIL introduces last-layer Mahalanobis distance and action chunk consistency detectors that together enable early, reliable failure detection in finetuned VLAs without failure data or expensive sampling.

Reference graph

Works this paper leans on

104 extracted references · 14 linked inside Pith · cited by 5 Pith papers

  1. [1]

    When to ask for help: Proactive interventions in autonomous reinforcement learning,

    A. Xie, F. Tajwar, A. Sharma, and C. Finn, “When to ask for help: Proactive interventions in autonomous reinforcement learning,” in Advances in Neural Information Processing Systems, 2022

    2022

  2. [2]

    Failure prediction with statistical guarantees for vision-based robot control,

    A. Farid, D. Snyder, A. Z. Ren, and A. Majumdar, “Failure prediction with statistical guarantees for vision-based robot control,” inRobotics: Science and Systems, 2022

    2022

  3. [3]

    Vision-language models as success detectors,

    Y . Du, K. Konyushkova, M. Denil, A. Raju, J. Landon, F. Hill, N. de Freitas, and S. Cabi, “Vision-language models as success detectors,” inConference on Lifelong Learning Agents, 2023

    2023

  4. [4]

    Asking for help: Failure prediction in behavioral cloning through value approximation,

    C. Gokmen, D. Ho, and M. Khansari, “Asking for help: Failure prediction in behavioral cloning through value approximation,” in International Conference on Robotics and Automation, 2023

    2023

  5. [5]

    Unpacking failure modes of generative policies: Runtime monitoring of consistency and progress,

    C. Agia, R. Sinha, J. Yang, Z. Cao, R. Antonova, M. Pavone, and J. Bohg, “Unpacking failure modes of generative policies: Runtime monitoring of consistency and progress,” inConference on Robot Learning, 2024

    2024

  6. [6]

    Grounding multimodal llms to embodied agents that ask for help with reinforcement learning,

    R. Ramrakhya, M. Chang, X. Puig, R. Desai, Z. Kira, and R. Mottaghi, “Grounding multimodal llms to embodied agents that ask for help with reinforcement learning,”arXiv preprint arXiv:2504.00907, 2025

    arXiv 2025

  7. [7]

    Collabvla: Self-reflective vision-language-action model dreaming together with human,

    N. Sun, Y . Li, C. Wang, H. Li, and H. Liu, “Collabvla: Self-reflective vision-language-action model dreaming together with human,”arXiv preprint arXiv:2509.14889, 2025

    arXiv 2025

  8. [8]

    SAFE: multitask failure detection for vision-language- action models,

    Q. Gu, Y . Ju, S. Sun, I. Gilitschenski, H. Nishimura, M. Itkina, and F. Shkurti, “SAFE: multitask failure detection for vision-language- action models,” inAdvances in Neural Information Processing Systems, 2025

    2025

  9. [9]

    REFLECT: summarizing robot experiences for failure explanation and correction,

    Z. Liu, A. Bahety, and S. Song, “REFLECT: summarizing robot experiences for failure explanation and correction,” inConference on Robot Learning, 2023

    2023

  10. [10]

    AHA: A vision- language-model for detecting and reasoning over failures in robotic manipulation,

    J. Duan, W. Pumacay, N. Kumar, Y . R. Wang, S. Tian, W. Yuan, R. Krishna, D. Fox, A. Mandlekar, and Y . Guo, “AHA: A vision- language-model for detecting and reasoning over failures in robotic manipulation,” inInternational Conference on Learning Representa- tions, 2025

    2025

  11. [11]

    KitchenVLA: Iterative vision-language corrections for robotic execution of human tasks,

    K. Lu, C. Ma, C. Hori, and D. Romeres, “KitchenVLA: Iterative vision-language corrections for robotic execution of human tasks,” in International Conference on Robotics and Automation Workshop on Safely Leveraging Vision-Language F oundation Models in Robotics, 2025

    2025

  12. [12]

    STAR: A foundation model-driven framework for robust task planning and failure recovery in robotic systems,

    M. S. Sakib and Y . Sun, “STAR: A foundation model-driven framework for robust task planning and failure recovery in robotic systems,”arXiv preprint arXiv:2503.06060, 2025

    Pith/arXiv arXiv 2025

  13. [13]

    Openvla: An open-source vision-language-action model,

    M. J. Kim, K. Pertsch, S. Karamcheti, T. Xiao, A. Balakrishna, S. Nair, R. Rafailov, E. P. Foster, P. R. Sanketi, Q. Vuong, T. Kollar, B. Burchfiel, R. Tedrake, D. Sadigh, S. Levine, P. Liang, and C. Finn, “Openvla: An open-source vision-language-action model,” inConference on Robot Learning, 2024

    2024

  14. [14]

    π0: A vision-language-action flow model for general robot control,

    K. Black, N. Brown, D. Driess, A. Esmail, M. Equi, C. Finn, N. Fusai, L. Groom, K. Hausman, B. Ichter, S. Jakubczak, T. Jones, L. Ke, S. Levine, A. Li-Bell, M. Mothukuri, S. Nair, K. Pertsch, L. X. Shi, J. Tanner, Q. Vuong, A. Walling, H. Wang, and U. Zhilinsky, “ π0: A vision-language-action flow model for general robot control,”arXiv preprint arXiv:2410...

    Pith/arXiv arXiv 2024

  15. [15]

    GR00T N1: an open foundation model for generalist humanoid robots,

    J. Bjorck, F. Castañeda, N. Cherniadev, X. Da, R. Ding, L. Fan, Y . Fang, D. Fox, F. Hu, S. Huang, J. Jang, Z. Jiang, J. Kautz, K. Kundalia, L. Lao, Z. Li, Z. Lin, K. Lin, G. Liu, E. LLontop, L. Magne, A. Mandlekar, A. Narayan, S. Nasiriany, S. Reed, Y . L. Tan, G. Wang, Z. Wang, J. Wang, Q. Wang, J. Xiang, Y . Xie, Y . Xu, Z. Xu, S. Ye, Z. Yu, A. Zhang, ...

    Pith/arXiv arXiv 2025

  16. [16]

    Policy adaptation via language optimization: Decomposing tasks for few-shot imitation,

    V . Myers, C. Zheng, O. Mees, K. Fang, and S. Levine, “Policy adaptation via language optimization: Decomposing tasks for few-shot imitation,” inConference on Robot Learning, 2024

    2024

  17. [17]

    π0.5: a vision-language-action model with open-world generalization,

    P. Intelligence, K. Black, N. Brown, J. Darpinian, K. Dhabalia, D. Driess, A. Esmail, M. Equi, C. Finn, N. Fusai, M. Y . Galliker, D. Ghosh, L. Groom, K. Hausman, B. Ichter, S. Jakubczak, T. Jones, L. Ke, D. LeBlanc, S. Levine, A. Li-Bell, M. Mothukuri, S. Nair, K. Pertsch, A. Z. Ren, L. X. Shi, L. Smith, J. T. Springenberg, K. Stachowicz, J. Tanner, Q. V...

    Pith/arXiv arXiv 2025

  18. [18]

    Hi robot: Open-ended instruction following with hierarchical vision-language-action models,

    L. X. Shi, B. Ichter, M. R. Equi, L. Ke, K. Pertsch, Q. Vuong, J. Tanner, A. Walling, H. Wang, N. Fusai, A. Li-Bell, D. Driess, L. Groom, S. Levine, and C. Finn, “Hi robot: Open-ended instruction following with hierarchical vision-language-action models,” inInternational Conference on Machine Learning, 2025

    2025

  19. [19]

    Seqvla: Sequential task execution for long-horizon manipulation with completion-aware vision- language-action model,

    R. Yang, Z. An, L. Zhou, and Y . Feng, “Seqvla: Sequential task execution for long-horizon manipulation with completion-aware vision- language-action model,”arXiv preprint arXiv:2509.14138, 2025

    arXiv 2025

  20. [20]

    Long-vla: Unleashing long-horizon capability of vision language action model for robot manipulation,

    Y . Fan, P. Ding, S. Bai, X. Tong, Y . Zhu, H. Lu, F. Dai, W. Zhao, Y . Liu, S. Huang, Z. Fan, B. Chen, and D. Wang, “Long-vla: Unleashing long-horizon capability of vision language action model for robot manipulation,” inConference on Robot Learning, 2025

    2025

  21. [21]

    Tactical rewind: Self-correction via backtracking in vision-and-language navigation,

    L. Ke, X. Li, Y . Bisk, A. Holtzman, Z. Gan, J. Liu, J. Gao, Y . Choi, and S. S. Srinivasa, “Tactical rewind: Self-correction via backtracking in vision-and-language navigation,” inConference on Computer Vision and Pattern Recognition, 2019

    2019

  22. [22]

    Smartway: Enhanced waypoint prediction and backtracking for zero- shot vision-and-language navigation,

    X. Shi, Z. Li, W. Lyu, J. Xia, F. Dayoub, Y . Qiao, and Q. Wu, “Smartway: Enhanced waypoint prediction and backtracking for zero- shot vision-and-language navigation,” inInternational Conference on Intelligent Robots and Systems, 2025

    2025

  23. [23]

    To err is robotic: Rapid value-based trial-and-error during deployment,

    M. Du, A. Khazatsky, T. Gerstenberg, and C. Finn, “To err is robotic: Rapid value-based trial-and-error during deployment,”arXiv preprint arXiv:2406.15917, 2024

    Pith/arXiv arXiv 2024

  24. [24]

    The llama 3 herd of models,

    T. L. . Team, “The llama 3 herd of models,”arXiv preprint arXiv:2407.21783, 2024

    Pith/arXiv arXiv 2024

  25. [25]

    Language models are few-shot learners,

    T. B. Brown, B. Mann, N. Ryder, M. Subbiah, J. Kaplan, P. Dhariwal, A. Neelakantan, P. Shyam, G. Sastry, A. Askell, S. Agarwal, A. Herbert- V oss, G. Krueger, T. Henighan, R. Child, A. Ramesh, D. M. Ziegler, J. Wu, C. Winter, C. Hesse, M. Chen, E. Sigler, M. Litwin, S. Gray, B. Chess, J. Clark, C. Berner, S. McCandlish, A. Radford, I. Sutskever, and D. Am...

    2020

  26. [26]

    Spatialpin: Enhancing spatial reasoning capabilities of vision-language models through prompting and interacting 3d priors,

    C. Ma, K. Lu, T.-Y . Cheng, N. Trigoni, and A. Markham, “Spatialpin: Enhancing spatial reasoning capabilities of vision-language models through prompting and interacting 3d priors,” inNeural Information Processing Systems, 2024

    2024

  27. [27]

    Touch and go: Learning from human-collected vision and touch,

    F. Yang, C. Ma, J. Zhang, J. Zhu, W. Yuan, and A. Owens, “Touch and go: Learning from human-collected vision and touch,”Advances in Neural Information Processing Systems, 2022

    2022

  28. [28]

    Improved baselines with visual instruction tuning,

    H. Liu, C. Li, Y . Li, and Y . J. Lee, “Improved baselines with visual instruction tuning,” inConference on Computer Vision and Pattern Recognition, 2024

    2024

  29. [29]

    Minimum bayes-risk decoding for statistical machine translation,

    S. Kumar and W. J. Byrne, “Minimum bayes-risk decoding for statistical machine translation,” inHuman Language Technology Conference of the North American Chapter of the Association for Computational Linguistics, 2004

    2004

  30. [30]

    Open x-embodiment: Robotic learning datasets and RT-X models : Open x-embodiment collaboration,

    O. X.-E. Collaboration, “Open x-embodiment: Robotic learning datasets and RT-X models : Open x-embodiment collaboration,” inInternational Conference on Robotics and Automation, 2024

    2024

  31. [31]

    RDT-1B: a diffusion foundation model for bimanual manip- ulation,

    S. Liu, L. Wu, B. Li, H. Tan, H. Chen, Z. Wang, K. Xu, H. Su, and J. Zhu, “RDT-1B: a diffusion foundation model for bimanual manip- ulation,” inInternational Conference on Learning Representations, 2025

    2025

  32. [32]

    Robomonkey: Scaling test-time sampling and veri- fication for vision-language-action models,

    J. Kwok, C. Agia, R. Sinha, M. Foutter, S. Li, I. Stoica, A. Mirhoseini, and M. Pavone, “Robomonkey: Scaling test-time sampling and veri- fication for vision-language-action models,” inConference on Robot Learning, 2025

    2025

  33. [33]

    Rover: Robot reward model as test-time verifier for vision- language-action model,

    M. Dai, L. Liu, Y . Bai, Y . Liu, Z. Wang, R. SU, C. Chen, L. Lin, and X. Wu, “Rover: Robot reward model as test-time verifier for vision- language-action model,”arXiv preprint arXiv:2510.10975, 2025

    arXiv 2025

  34. [34]

    Sampling-based approximations to minimum bayes risk decoding for neural machine translation,

    B. Eikema and W. Aziz, “Sampling-based approximations to minimum bayes risk decoding for neural machine translation,” inEmpirical Methods in Natural Language Processing, 2022

    2022

  35. [35]

    On extending direct preference optimization to accommodate ties,

    J. Chen, G. Yang, W. Lin, J. Mei, and B. Byrne, “On extending direct preference optimization to accommodate ties,” inAdvances in Neural Information Processing Systems, 2025

    2025

  36. [36]

    Direct preference optimization for neural machine translation with minimum bayes risk decoding,

    G. Yang, J. Chen, W. Lin, and B. Byrne, “Direct preference optimization for neural machine translation with minimum bayes risk decoding,” inConference of the North American Chapter of the Association for Computational Linguistics, 2024

    2024

  37. [37]

    Can we detect failures without failure data? uncertainty-aware runtime failure detection for imitation learning policies,

    C. Xu, T. K. Nguyen, E. Dixon, C. Rodriguez, P. T. Miller, R. Lee, P. Shah, R. Ambrus, H. Nishimura, and M. Itkina, “Can we detect failures without failure data? uncertainty-aware runtime failure detection for imitation learning policies,”arXiv preprint arXiv:2503.08558, 2025

    Pith/arXiv arXiv 2025

  38. [38]

    Sparse and complete latent organization for geospatial semantic segmentation,

    F. Yang and C. Ma, “Sparse and complete latent organization for geospatial semantic segmentation,” inConference on Computer Vision and Pattern Recognition, 2022

    2022

  39. [39]

    Error-aware imitation learning from teleopera- tion data for mobile manipulation,

    J. Wong, A. Tung, A. Kurenkov, A. Mandlekar, L. Fei-Fei, S. Savarese, and R. Martín-Martín, “Error-aware imitation learning from teleopera- tion data for mobile manipulation,” inConference on Robot Learning, 2021

    2021

  40. [40]

    Model-based runtime monitoring with interactive imitation learning,

    H. Liu, S. Dass, R. Martín-Martín, and Y . Zhu, “Model-based runtime monitoring with interactive imitation learning,” inInternational Conference on Robotics and Automation, 2024

    2024

  41. [41]

    Real-time anomaly detection and reactive planning with large language models,

    R. Sinha, A. Elhafsi, C. Agia, M. Foutter, E. Schmerling, and M. Pavone, “Real-time anomaly detection and reactive planning with large language models,” inRobotics: Science and Systems, 2024

    2024

  42. [42]

    Doremi: Grounding language model by detecting and recovering from plan-execution misalignment,

    Y . Guo, Y . Wang, L. Zha, and J. Chen, “Doremi: Grounding language model by detecting and recovering from plan-execution misalignment,” inInternational Conference on Intelligent Robots and Systems, 2024

    2024

  43. [43]

    A self-correcting vision-language- action model for fast and slow system manipulation,

    C. Li, J. Liu, G. Wang, X. Li, S. Chen, L. Heng, C. Xiong, J. Ge, R. Zhang, K. Zhou, and S. Zhang, “A self-correcting vision-language- action model for fast and slow system manipulation,”arXiv preprint arXiv:2405.17418, 2025

    Pith/arXiv arXiv 2025

  44. [44]

    Coopera: Continual open-ended human-robot assistance,

    C. Ma, K. Lu, R. Desai, X. Puig, A. Markham, and N. Trigoni, “Coopera: Continual open-ended human-robot assistance,” inAdvances in Neural Information Processing Systems, 2025

    2025

  45. [45]

    Multi-task interactive robot fleet learning with visual world models,

    H. Liu, Y . Zhang, V . Betala, E. Zhang, J. Liu, C. Ding, and Y . Zhu, “Multi-task interactive robot fleet learning with visual world models,” inConference on Robot Learning, 2024

    2024

  46. [46]

    Cot-vla: Visual chain-of-thought reasoning for vision-language-action models,

    Q. Zhao, Y . Lu, M. J. Kim, Z. Fu, Z. Zhang, Y . Wu, Z. Li, Q. Ma, S. Han, C. Finn, A. Handa, T. Lin, G. Wetzstein, M. Liu, and D. Xiang, “Cot-vla: Visual chain-of-thought reasoning for vision-language-action models,” inConference on Computer Vision and Pattern Recognition, 2025

    2025

  47. [47]

    Reflective planning: Vision-language models for multi-stage long-horizon robotic manipulation,

    Y . Feng, J. Han, Z. Yang, X. Yue, S. Levine, and J. Luo, “Reflective planning: Vision-language models for multi-stage long-horizon robotic manipulation,”arXiv preprint arXiv:2502.16707, 2025

    Pith/arXiv arXiv 2025

  48. [48]

    V-JEPA 2: Self-supervised video models enable understanding, prediction and planning,

    M. Assran, A. Bardes, D. Fan, Q. Garrido, R. Howes, M. Komeili, M. J. Muckley, A. Rizvi, C. Roberts, K. Sinha, A. Zholus, S. Arnaud, A. Gejji, A. Martin, F. R. Hogan, D. Dugas, P. Bojanowski, V . Khalidov, P. Labatut, F. Massa, M. Szafraniec, K. Krishnakumar, Y . Li, X. Ma, S. Chandar, F. Meier, Y . LeCun, M. Rabbat, and N. Ballas, “V-JEPA 2: Self-supervi...

    Pith/arXiv arXiv 2025

  49. [49]

    Dexvla: Vision- language model with plug-in diffusion expert for general robot control,

    J. Wen, Y . Zhu, J. Li, Z. Tang, C. Shen, and F. Feng, “Dexvla: Vision- language model with plug-in diffusion expert for general robot control,” inConference on Robot Learning, 2025

    2025

  50. [50]

    RDD: retrieval-based demon- stration decomposer for planner alignment in long-horizon tasks,

    M. Yan, Y . Wang, Z. Liu, and J. Li, “RDD: retrieval-based demon- stration decomposer for planner alignment in long-horizon tasks,” in Advances in Neural Information Processing Systems, 2025

    2025

  51. [51]

    Robotic control via embodied chain-of-thought reasoning,

    M. Zawalski, W. Chen, K. Pertsch, O. Mees, C. Finn, and S. Levine, “Robotic control via embodied chain-of-thought reasoning,” inConfer- ence on Robot Learning, 2024

    2024

  52. [52]

    Chatvla: Unified multimodal understanding and robot control with vision-language-action model,

    Z. Zhou, Y . Zhu, M. Zhu, J. Wen, N. Liu, Z. Xu, W. Meng, R. Cheng, Y . Peng, C. Shen, and F. Feng, “Chatvla: Unified multimodal understanding and robot control with vision-language-action model,” inEmpirical Methods in Natural Language Processing, 2025

    2025

  53. [53]

    Chatvla-2: Vision- language-action model with open-world embodied reasoning from pretrained knowledge,

    Z. Zhou, Y . Zhu, J. Wen, C. Shen, and Y . Xu, “Chatvla-2: Vision- language-action model with open-world embodied reasoning from pretrained knowledge,” inAdvances in Neural Information Processing Systems, 2025

    2025

  54. [54]

    Training strategies for efficient embodied reasoning,

    W. Chen, S. Belkhale, S. Mirchandani, O. Mees, D. Driess, K. Pertsch, and S. Levine, “Training strategies for efficient embodied reasoning,” inConference on Robot Learning, 2025

    2025

  55. [55]

    World simulation with video foundation models for physical ai,

    NVIDIA, “World simulation with video foundation models for physical ai,”arXiv preprint arXiv:2511.00062, 2025

    Pith/arXiv arXiv 2025

  56. [56]

    Run-time observation interventions make vision-language-action models more visually robust,

    A. J. Hancock, A. Z. Ren, and A. Majumdar, “Run-time observation interventions make vision-language-action models more visually robust,” inInternational Conference on Robotics and Automation, 2025

    2025

  57. [57]

    IA-VLA: input augmentation for vision-language-action models in settings with semantically complex tasks,

    E. Hannus, M. Malin, T. N. Le, and V . Kyrki, “IA-VLA: input augmentation for vision-language-action models in settings with semantically complex tasks,”arXiv preprint arXiv:2509.24768, 2025

    arXiv 2025

  58. [58]

    Visual prompting via image inpainting,

    A. Bar, Y . Gandelsman, T. Darrell, A. Globerson, and A. A. Efros, “Visual prompting via image inpainting,” inAdvances in Neural Information Processing Systems, 2022

    2022

  59. [59]

    Scaling LLM test-time compute optimally can be more effective than scaling model parameters,

    C. Snell, J. Lee, K. Xu, and A. Kumar, “Scaling LLM test-time compute optimally can be more effective than scaling model parameters,” in International Conference on Learning Representations, 2025

    2025

  60. [60]

    Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning,

    DeepSeek-AI, “Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning,”arXiv preprint arXiv:2501.12948, 2025

    Pith/arXiv arXiv 2025

  61. [61]

    Introducing openai o3 and o4-mini,

    O. AI, “Introducing openai o3 and o4-mini,” https://openai.com/index/ introducing-o3-and-o4-mini/, 2025

    2025

  62. [62]

    Steering your generalists: Improving robotic foundation models via value guidance,

    M. Nakamoto, O. Mees, A. Kumar, and S. Levine, “Steering your generalists: Improving robotic foundation models via value guidance,” inConference on Robot Learning, 2024

    2024

  63. [63]

    Navid: Video-based VLM plans the next step for vision-and-language navigation,

    J. Zhang, K. Wang, R. Xu, G. Zhou, Y . Hong, X. Fang, Q. Wu, Z. Zhang, and H. Wang, “Navid: Video-based VLM plans the next step for vision-and-language navigation,” inRobotics: Science and Systems, 2024

    2024

  64. [64]

    Chain-of-thought prompting elicits reasoning in large language models,

    J. Wei, X. Wang, D. Schuurmans, M. Bosma, B. Ichter, F. Xia, E. H. Chi, Q. V . Le, and D. Zhou, “Chain-of-thought prompting elicits reasoning in large language models,” inAdvances in Neural Information Processing Systems, 2022

    2022

  65. [65]

    Introducing gpt-4.1 in the api,

    O. AI, “Introducing gpt-4.1 in the api,” https://openai.com/index/ gpt-4-1/, 2025

    2025

  66. [66]

    Fine-tuning vision-language-action models: Optimizing speed and success,

    M. J. Kim, C. Finn, and P. Liang, “Fine-tuning vision-language-action models: Optimizing speed and success,” inRobotics: Science and Systems, 2025

    2025

  67. [67]

    Introducing gpt-5.2,

    O. AI, “Introducing gpt-5.2,” https://openai.com/index/ introducing-gpt-5-2/, 2025

    2025

  68. [68]

    LIBERO: benchmarking knowledge transfer for lifelong robot learning,

    B. Liu, Y . Zhu, C. Gao, Y . Feng, Q. Liu, Y . Zhu, and P. Stone, “LIBERO: benchmarking knowledge transfer for lifelong robot learning,” inAdvances in Neural Information Processing Systems, 2023

    2023

  69. [69]

    Tracevla: Visual trace prompting enhances spatial- temporal awareness for generalist robotic policies,

    R. Zheng, Y . Liang, S. Huang, J. Gao, H. D. III, A. Kolobov, F. Huang, and J. Yang, “Tracevla: Visual trace prompting enhances spatial- temporal awareness for generalist robotic policies,” inInternational Conference on Learning Representations, 2025

    2025

  70. [70]

    Thinkact: Vision-language-action reasoning via reinforced visual latent planning,

    C.-P. Huang, Y .-H. Wu, M.-H. Chen, Y .-C. F. Wang, and F.-E. Yang, “Thinkact: Vision-language-action reasoning via reinforced visual latent planning,” inAdvances in Neural Information Processing Systems, 2025

    2025

  71. [71]

    FPC-VLA: A vision-language-action framework with a supervisor for failure prediction and correction,

    Y . Yang, Z. Duan, T. Xie, F. Cao, P. Shen, P. Song, P. Jin, G. Sun, S. Xu, Y . You, and J. Liu, “FPC-VLA: A vision-language-action framework with a supervisor for failure prediction and correction,”arXiv preprint arXiv:2509.04018, 2025

    arXiv 2025

  72. [72]

    Cogvla: Cognition- aligned vision-language-action model via instruction-driven routing & sparsification,

    W. Li, R. Zhang, R. Shao, J. He, and L. Nie, “Cogvla: Cognition- aligned vision-language-action model via instruction-driven routing & sparsification,” inAdvances in Neural Information Processing Systems, 2025

    2025

  73. [73]

    Diffusion policy: Visuomotor policy learning via action diffusion,

    C. Chi, S. Feng, Y . Du, Z. Xu, E. Cousineau, B. Burchfiel, and S. Song, “Diffusion policy: Visuomotor policy learning via action diffusion,” in Robotics: Science and Systems, 2023

    2023

  74. [74]

    Octo: An open-source generalist robot policy,

    D. Ghosh, H. R. Walke, K. Pertsch, K. Black, O. Mees, S. Dasari, J. Hejna, T. Kreiman, C. Xu, J. Luo, Y . L. Tan, L. Y . Chen, Q. Vuong, T. Xiao, P. R. Sanketi, D. Sadigh, C. Finn, and S. Levine, “Octo: An open-source generalist robot policy,” inRobotics: Science and Systems, 2024

    2024

  75. [75]

    Spatialvla: Exploring spatial representations for visual-language-action model,

    D. Qu, H. Song, Q. Chen, Y . Yao, X. Ye, Y . Ding, Z. Wang, J. Gu, B. Zhao, D. Wanget al., “Spatialvla: Exploring spatial representations for visual-language-action model,”arXiv preprint arXiv:2501.15830, 2025

    Pith/arXiv arXiv 2025

  76. [76]

    Improving minimum bayes risk decoding with multi-prompt,

    D. Heineman, Y . Dou, and W. Xu, “Improving minimum bayes risk decoding with multi-prompt,” inEmpirical Methods in Natural Language Processing, 2024

    2024

  77. [77]

    Fast best-of-n decoding via speculative rejection,

    H. Sun, M. Haider, R. Zhang, H. Yang, J. Qiu, M. Yin, M. Wang, P. L. Bartlett, and A. Zanette, “Fast best-of-n decoding via speculative rejection,” inAdvances in Neural Information Processing Systems, 2024

    2024

  78. [78]

    Towards understanding sycophancy in language models,

    M. Sharma, M. Tong, T. Korbak, D. Duvenaud, A. Askell, S. R. Bowman, E. Durmus, Z. Hatfield-Dodds, S. R. Johnston, S. Kravec, T. Maxwell, S. McCandlish, K. Ndousse, O. Rausch, N. Schiefer, D. Yan, M. Zhang, and E. Perez, “Towards understanding sycophancy in language models,” inInternational Conference on Learning Repre- sentations, 2024

    2024

  79. [79]

    Efficient vertical federated learning with secure aggregation,

    X. Qiu, H. Pan, W. Zhao, C. Ma, P. P. B. de Gusmao, and N. D. Lane, “Efficient vertical federated learning with secure aggregation,” in Federated Learning Systems (FLSys) Workshop@ MLSys 2023, 2023

    2023

  80. [80]

    vfedsec: Efficient secure aggregation for vertical federated learning via secure layer,

    X. Qiu, H. Pan, W. Zhao, C. Ma, P. P. Gusmao, and N. D. Lane, “vfedsec: Efficient secure aggregation for vertical federated learning via secure layer,”arXiv preprint arXiv:2305.16794, 2023

    Pith/arXiv arXiv 2023

Showing first 80 references.

This paper was first reviewed by deepseek-v4-flash on August 3, 2026.