pith. sign in

arxiv: 2607.02195 · v1 · pith:KZWWJDEGnew · submitted 2026-07-02 · 💻 cs.RO

Bridge-WA: Predicting Where and How the World Changes for Robotic Action

Yongjie Bai , Hanting Wang , Mingtong Dai , Qijun Zhong , Yang Liu , Liang Lin This is my paper

Pith reviewed 2026-07-03 11:26 UTC · model grok-4.3

classification 💻 cs.RO
keywords robotic manipulationworld-action modelsscene change predictionmodel distillationaction generalizationvision-language-actionfuture priors
0
0 comments X

The pith

Bridge-WA distills scene-change predictions into compact priors that condition robotic actions and improve generalization without running a full generative world model at deployment.

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

The paper tries to establish that robotic action policies can be made more effective and robust by focusing only on where and how the scene will change rather than predicting complete future images. It does so by distilling a teacher model into three lightweight priors that capture intended outcomes, intervention locations, and motion directions, then feeding those priors into the action generator through a conditioning module. This setup removes the expensive teacher at test time while suppressing irrelevant visual details such as backgrounds and lighting variations. A reader would care because it offers a way to retain the benefits of world models for control without their full computational burden during robot operation.

Core claim

Bridge-WA distills a frozen future-change teacher into three compact priors—future tokens for intended outcomes, change maps for intervention support, and motion-flow maps for local transition direction—then conditions an action transformer on these priors through multi-source attention memories and spatial-temporal biases inside the WorldBridge module, allowing the teacher to be removed at inference while raising task success, progress, and robustness across VLABench, RoboTwin2.0, LIBERO-Plus, and real-robot tests, especially under out-of-distribution visual shifts.

What carries the argument

The WorldBridge module, which transfers the three distilled change priors to the action transformer using multi-source attention memories and spatial-temporal biases.

If this is right

  • Task success and progress increase on the listed simulation and real-robot benchmarks.
  • Robustness improves under shifts in background, lighting, and distractors.
  • Deployment no longer requires running dense future-image generation.
  • Nuisance appearance factors are suppressed because attention is directed to change locations and directions.

Where Pith is reading between the lines

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

  • The same change priors could be attached to other action architectures beyond the tested transformer.
  • The method may reduce memory and compute needs enough for onboard robot deployment.
  • Extending the priors to longer time horizons could be tested by measuring performance drop on multi-step tasks.

Load-bearing premise

The three distilled priors plus the attention memories are enough to carry all action-relevant information from the teacher without meaningful loss.

What would settle it

An evaluation on a manipulation task with strong visual distractors where the Bridge-WA policy achieves lower success than a baseline that keeps the full future-image teacher active at inference.

Figures

Figures reproduced from arXiv: 2607.02195 by Hanting Wang, Liang Lin, Mingtong Dai, Qijun Zhong, Yang Liu, Yongjie Bai.

Figure 1
Figure 1. Figure 1: BRIDGE-WA represents the action-relevant future with three compact world-change repre￾sentations: future tokens for the intended outcome, change maps for where the scene should change, and motion-flow maps for how the change should move. A lightweight predictor estimates these rep￾resentations from the current robot context, and WORLDBRIDGE conditions the action transformer on them for world-aware action g… view at source ↗
Figure 2
Figure 2. Figure 2: (I) Robot-state-conditioned teacher world model based on the Wan2.2-5B generative back [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Real-robot visualization analysis. (I) Five manipulation tasks and OOD evaluation scenes, [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Experimental results across simulation, ablation, and real-robot evaluations. (I)-(III) [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Dobot and Franka real-world setup. Overview. We evaluate the real-robot task execution capability and OOD generalization capabil￾ity of BRIDGE-WA on two single-arm platforms: a Dobot Nova2 robot and a Franka FR3 robot. Both platforms use LeRobot-style data with synchronized third-person RGB, wrist RGB, Cartesian robot state, Cartesian end-effector action, timestamps, and task instructions. The action/state… view at source ↗
Figure 6
Figure 6. Figure 6: Task and scene visualizations for the simulation benchmarks. The figure summarizes [PITH_FULL_IMAGE:figures/full_fig_p020_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Qualitative world-prior visualizations on RoboTwin 2.0 (I) and VLABench (II). We show [PITH_FULL_IMAGE:figures/full_fig_p020_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Dobot real-robot execution sequences. The figure shows temporal rollouts for the five [PITH_FULL_IMAGE:figures/full_fig_p021_8.png] view at source ↗
read the original abstract

General-purpose vision-language-action models benefit from large vision-language priors, but effective manipulation also requires anticipating action-relevant scene changes. Existing world-action models often rely on large generative world models or dense future rollouts, which are expensive and spend capacity on visual details weakly coupled to control. We present Bridge-WA, a lightweight world-action framework that distills a frozen future-change teacher into three compact priors: future tokens for intended outcomes, change maps for intervention support, and motion-flow maps for local transition direction. A WorldBridge conditions the action transformer on these priors through multi-source attention memories and spatial-temporal biases, while the teacher model is removed at inference. Across VLABench, RoboTwin2.0, LIBERO-Plus and real-robot evaluations, Bridge-WA improves task success, progress, and robustness, with particularly clear gains under out-of-distribution visual shifts. By focusing action generation on where and how the scene will change, Bridge-WA suppresses nuisance appearance factors such as background, lighting, and distractors, leading to better generalization without deployment-time dense future-image generation. Code and visualizations are available at: https://hcplab-sysu.github.io/BRIDGE-WA .

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The paper presents Bridge-WA, a lightweight world-action framework for vision-language-action models that distills a frozen future-change teacher into three compact priors (future tokens for intended outcomes, change maps for intervention support, and motion-flow maps for local transition direction). A WorldBridge module conditions the action transformer on these priors via multi-source attention memories and spatial-temporal biases, with the teacher removed at inference. The central claim is that focusing on where and how the scene changes suppresses nuisance factors (background, lighting, distractors) and yields better generalization and robustness on VLABench, RoboTwin2.0, LIBERO-Plus, and real-robot tasks without requiring deployment-time dense future-image generation.

Significance. If the empirical gains and lack of performance penalty hold, the work would be significant for efficient VLA robotics by replacing expensive generative world models with targeted change priors, potentially improving OOD robustness while lowering inference cost. The explicit distillation of change-focused signals and removal of the teacher at test time are concrete engineering contributions that could influence subsequent world-action architectures.

major comments (2)
  1. [Method section (distillation and conditioning)] Method (distillation and WorldBridge conditioning): The claim that the three distilled priors plus multi-source attention memories transfer all action-relevant teacher knowledge without significant loss is load-bearing for both the generalization benefit and the 'no performance penalty' assertion. The manuscript should provide ablations that isolate each prior (future tokens, change maps, motion-flow maps) and quantify any drop in task success or progress relative to the full teacher, particularly for multi-step dependencies and contact dynamics.
  2. [Experiments section (evaluation tables)] Experiments (evaluation tables): The abstract and results claim clear gains under visual shifts, but the central claim of effective nuisance suppression rests on the priors capturing semantically relevant intervention effects. Tables reporting task success, progress, and robustness should include per-component ablations and direct comparisons against the frozen teacher to confirm no critical signal is lost in distillation.
minor comments (1)
  1. [Abstract] The abstract states improvements across four benchmarks but does not report specific quantitative deltas, baseline names, or ablation details; these should be summarized with numbers in the abstract for clarity.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback. We address each major comment below and will revise the manuscript accordingly.

read point-by-point responses

  1. Referee: [Method section (distillation and conditioning)] Method (distillation and WorldBridge conditioning): The claim that the three distilled priors plus multi-source attention memories transfer all action-relevant teacher knowledge without significant loss is load-bearing for both the generalization benefit and the 'no performance penalty' assertion. The manuscript should provide ablations that isolate each prior (future tokens, change maps, motion-flow maps) and quantify any drop in task success or progress relative to the full teacher, particularly for multi-step dependencies and contact dynamics.

    Authors: We agree that component-wise ablations are needed to substantiate the distillation claim. The current manuscript reports overall gains from the full set of priors but does not isolate each prior or provide direct per-task comparisons to the frozen teacher on multi-step or contact-rich scenarios. In the revised version we will add these ablations, reporting task success and progress metrics for each prior in isolation and in combination against the teacher. revision: yes

  2. Referee: [Experiments section (evaluation tables)] Experiments (evaluation tables): The abstract and results claim clear gains under visual shifts, but the central claim of effective nuisance suppression rests on the priors capturing semantically relevant intervention effects. Tables reporting task success, progress, and robustness should include per-component ablations and direct comparisons against the frozen teacher to confirm no critical signal is lost in distillation.

    Authors: We concur that the evaluation tables should be expanded to include per-component ablations and explicit teacher comparisons. While the existing results demonstrate improved robustness without the teacher at inference, the tables do not currently break down contributions per prior or show direct teacher baselines on the reported metrics. We will update the tables in the revision to incorporate these analyses across VLABench, RoboTwin2.0, LIBERO-Plus, and real-robot tasks. revision: yes

Circularity Check

0 steps flagged

No circularity; empirical distillation framework with independent evaluation claims

full rationale

The paper describes an engineering framework that distills a frozen teacher into three priors (future tokens, change maps, motion-flow maps) and conditions an action transformer via WorldBridge. No equations, fitted parameters, or self-citations are presented in the provided text that reduce any claimed prediction or uniqueness result to the inputs by construction. The central assertion—that the priors capture action-relevant information while suppressing nuisance factors—is framed as an empirical hypothesis tested on VLABench, RoboTwin2.0, LIBERO-Plus and real-robot tasks, not as a self-definitional or self-citation-dependent derivation. This is a standard non-circular empirical contribution.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review; no explicit free parameters, axioms, or invented entities are stated or derivable from the provided text.

pith-pipeline@v0.9.1-grok · 5751 in / 1081 out tokens · 26386 ms · 2026-07-03T11:26:44.775005+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

67 extracted references · 26 canonical work pages · 23 internal anchors

  1. [1]

    M. Ahn, A. Brohan, N. Brown, Y . Chebotar, O. Cortes, B. David, C. Finn, C. Fu, K. Gopalakr- ishnan, K. Hausman, et al. Do as i can, not as i say: Grounding language in robotic affordances. arXiv preprint arXiv:2204.01691, 2022

    work page internal anchor Pith review Pith/arXiv arXiv 2022

  2. [2]

    Huang, F

    W. Huang, F. Xia, T. Xiao, H. Chan, J. Liang, P. Florence, A. Zeng, J. Tompson, I. Mordatch, Y . Chebotar, et al. Inner monologue: Embodied reasoning through planning with language models. InConference on Robot Learning, pages 1769–1782. PMLR, 2023

    2023

  3. [3]

    Liang, W

    J. Liang, W. Huang, F. Xia, P. Xu, K. Hausman, B. Ichter, P. Florence, and A. Zeng. Code as policies: Language model programs for embodied control. In2023 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2023

    2023

  4. [4]

    Driess, F

    D. Driess, F. Xia, M. S. Sajjadi, C. Lynch, A. Chowdhery, B. Ichter, A. Wahid, J. Tompson, Q. Vuong, T. Yu, et al. Palm-e: an embodied multimodal language model. InProceedings of the 40th International Conference on Machine Learning, pages 8469–8488, 2023

    2023

  5. [5]

    S. Reed, K. Zolna, E. Parisotto, S. G. Colmenarejo, A. Novikov, G. Barth-maron, M. Gim´enez, Y . Sulsky, J. Kay, J. T. Springenberg, et al. A generalist agent.Transactions on Machine Learning Research

  6. [6]

    Brohan, N

    A. Brohan, N. Brown, J. Carbajal, Y . Chebotar, J. Dabis, C. Finn, K. Gopalakrishnan, K. Haus- man, A. Herzog, J. Hsu, et al. Rt-1: Robotics transformer for real-world control at scale. Robotics: Science and Systems XIX, 2023

    2023

  7. [7]

    Zitkovich, T

    B. Zitkovich, T. Yu, S. Xu, P. Xu, T. Xiao, F. Xia, J. Wu, P. Wohlhart, S. Welker, A. Wahid, et al. Rt-2: Vision-language-action models transfer web knowledge to robotic control. In Conference on Robot Learning, pages 2165–2183. PMLR, 2023

    2023

  8. [8]

    O’Neill, A

    A. O’Neill, A. Rehman, A. Maddukuri, A. Gupta, A. Padalkar, A. Lee, A. Pooley, A. Gupta, A. Mandlekar, A. Jain, et al. Open x-embodiment: Robotic learning datasets and rt-x models: Open x-embodiment collaboration 0. In2024 IEEE International Conference on Robotics and Automation (ICRA), pages 6892–6903. IEEE, 2024

    2024

  9. [9]

    M. J. Kim, K. Pertsch, S. Karamcheti, T. Xiao, A. Balakrishna, S. Nair, R. Rafailov, E. P. Foster, P. R. Sanketi, Q. Vuong, et al. Openvla: An open-source vision-language-action model. In8th Annual Conference on Robot Learning

  10. [10]

    O. M. Team, D. Ghosh, H. Walke, K. Pertsch, K. Black, O. Mees, S. Dasari, J. Hejna, T. Kreiman, C. Xu, et al. Octo: An open-source generalist robot policy.arXiv preprint arXiv:2405.12213, 2024

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  11. [11]

    $\pi_0$: 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, et al. Pi0: A vision-language-action flow model for general robot control.arXiv preprint arXiv:2410.24164, 2024

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  12. [12]

    Florence, C

    P. Florence, C. Lynch, A. Zeng, O. A. Ramirez, A. Wahid, L. Downs, A. Wong, J. Lee, I. Mor- datch, and J. Tompson. Implicit behavioral cloning. InConference on robot learning, pages 158–168. PMLR, 2022

    2022

  13. [13]

    N. M. Shafiullah, Z. Cui, A. A. Altanzaya, and L. Pinto. Behavior transformers: Cloningk modes with one stone.Advances in neural information processing systems, 35:22955–22968, 2022

    2022

  14. [14]

    T. Z. Zhao, V . Kumar, S. Levine, and C. Finn. Learning fine-grained bimanual manipulation with low-cost hardware.arXiv preprint arXiv:2304.13705, 2023. 10

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  15. [15]

    C. Chi, Z. Xu, S. Feng, E. Cousineau, Y . Du, B. Burchfiel, R. Tedrake, and S. Song. Diffusion policy: Visuomotor policy learning via action diffusion.The International Journal of Robotics Research, 44(10-11):1684–1704, 2025

    2025

  16. [16]

    World Models

    D. Ha and J. Schmidhuber. World models.arXiv preprint arXiv:1803.10122, 2018

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  17. [17]

    Hafner, T

    D. Hafner, T. Lillicrap, I. Fischer, R. Villegas, D. Ha, H. Lee, and J. Davidson. Learning latent dynamics for planning from pixels. InInternational conference on machine learning, pages 2555–2565. PMLR, 2019

    2019

  18. [18]

    Hafner, T

    D. Hafner, T. Lillicrap, J. Ba, and M. Norouzi. Dream to control: Learning behaviors by latent imagination. InInternational Conference on Learning Representations, 2020

    2020

  19. [19]

    Mastering Diverse Domains through World Models

    D. Hafner, J. Pasukonis, J. Ba, and T. Lillicrap. Mastering diverse domains through world models.arXiv preprint arXiv:2301.04104, 2023

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  20. [20]

    Schrittwieser, I

    J. Schrittwieser, I. Antonoglou, T. Hubert, K. Simonyan, L. Sifre, S. Schmitt, A. Guez, E. Lock- hart, D. Hassabis, T. Graepel, et al. Mastering atari, go, chess and shogi by planning with a learned model.Nature, 588(7839):604–609, 2020

    2020

  21. [21]

    Finn and S

    C. Finn and S. Levine. Deep visual foresight for planning robot motion. In2017 IEEE inter- national conference on robotics and automation (ICRA), pages 2786–2793. IEEE, 2017

    2017

  22. [22]

    Visual Foresight: Model-Based Deep Reinforcement Learning for Vision-Based Robotic Control

    F. Ebert, C. Finn, S. Dasari, A. Xie, A. Lee, and S. Levine. Visual foresight: Model-based deep reinforcement learning for vision-based robotic control.arXiv preprint arXiv:1812.00568, 2018

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  23. [23]

    Babaeizadeh, M

    M. Babaeizadeh, M. T. Saffar, S. Nair, S. Levine, C. Finn, and D. Erhan. Fitvid: Overfitting in pixel-level video prediction.arXiv preprint arXiv:2106.13195, 2021

    work page arXiv 2021

  24. [24]

    Y . Du, S. Yang, B. Dai, H. Dai, O. Nachum, J. Tenenbaum, D. Schuurmans, and P. Abbeel. Learning universal policies via text-guided video generation.Advances in neural information processing systems, 36:9156–9172, 2023

    2023

  25. [25]

    S. Zhou, Y . Du, J. Chen, Y . Li, D.-Y . Yeung, and C. Gan. Robodreamer: Learning composi- tional world models for robot imagination. InInternational Conference on Machine Learning, pages 61885–61896. PMLR, 2024

    2024

  26. [26]

    B. Hou, G. Li, J. Jia, T. An, X. Guo, S. Leng, H. Geng, Y . Ze, T. Harada, P. Torr, et al. World model for robot learning: A comprehensive survey.arXiv preprint arXiv:2605.00080, 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  27. [27]

    S. Wang, J. Shi, Z. Fu, X. He, F. Liu, C. Yang, Y . Zhou, Z. Fei, J. Gong, J. Fu, et al. World action models: The next frontier in embodied ai.arXiv preprint arXiv:2605.12090, 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  28. [28]

    H. Luo, W. Zhang, Y . Feng, S. Zheng, H. Xu, C. Xu, Z. Xi, Y . Fu, and Z. Lu. Being-h0. 7: A latent world-action model from egocentric videos.arXiv preprint arXiv:2605.00078, 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  29. [29]

    Y . Liu, P. Sun, S. Li, Y . Xie, L. Zhang, X. Chao, S. Dong, F. Chen, X.-P. Zhang, and W. Ding. Oa-wam: Object-addressable world action model for robust robot manipulation.arXiv preprint arXiv:2605.06481, 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  30. [30]

    S. Nair, A. Rajeswaran, V . Kumar, C. Finn, and A. Gupta. R3m: A universal visual repre- sentation for robot manipulation. InConference on Robot Learning, pages 892–909. PMLR, 2023

    2023

  31. [31]

    Y . J. Ma, S. Sodhani, D. Jayaraman, O. Bastani, V . Kumar, and A. Zhang. Vip: Towards universal visual reward and representation via value-implicit pre-training. InThe Eleventh International Conference on Learning Representations, 2023. 11

    2023

  32. [32]

    Karamcheti, S

    S. Karamcheti, S. Nair, A. S. Chen, T. Kollar, C. Finn, D. Sadigh, and P. Liang. Language- driven representation learning for robotics.Robotics: Science and Systems, 2023

    2023

  33. [33]

    Hendrycks and T

    D. Hendrycks and T. Dietterich. Benchmarking neural network robustness to common corrup- tions and perturbations. InInternational Conference on Learning Representations, 2019

    2019

  34. [34]

    Hendrycks, S

    D. Hendrycks, S. Basart, N. Mu, S. Kadavath, F. Wang, E. Dorundo, R. Desai, T. Zhu, S. Para- juli, M. Guo, et al. The many faces of robustness: A critical analysis of out-of-distribution generalization. In2021 IEEE/CVF International Conference on Computer Vision (ICCV), pages 8320–8329. IEEE, 2021

    2021

  35. [35]

    Zheng, J

    J. Zheng, J. Li, Z. Wang, D. Liu, X. Kang, Y . Feng, Y . Zheng, J. Zou, Y . Chen, J. Zeng, et al. X- vla: Soft-prompted transformer as scalable cross-embodiment vision-language-action model. International Conference on Learning Representation, 2026

    2026

  36. [36]

    E. Jang, A. Irpan, M. Khansari, D. Kappler, F. Ebert, C. Lynch, S. Levine, and C. Finn. Bc-z: Zero-shot task generalization with robotic imitation learning. In5th Annual Conference on Robot Learning

  37. [37]

    Jiang, A

    Y . Jiang, A. Gupta, Z. Zhang, G. Wang, Y . Dou, Y . Chen, L. Fei-Fei, A. Anandkumar, Y . Zhu, and L. Fan. Vima: Robot manipulation with multimodal prompts. InInternational Conference on Machine Learning, pages 14975–15022. PMLR, 2023

    2023

  38. [38]

    Bousmalis, G

    K. Bousmalis, G. Vezzani, D. Rao, C. M. Devin, A. X. Lee, M. B. Villalonga, T. Davchev, Y . Zhou, A. Gupta, A. Raju, et al. Robocat: A self-improving generalist agent for robotic manipulation.Transactions on Machine Learning Research

  39. [39]

    Black, N

    K. Black, N. Brown, J. Darpinian, K. Dhabalia, D. Driess, A. Esmail, M. R. Equi, C. Finn, N. Fusai, M. Y . Galliker, et al. Pi0.5: a vision-language-action model with open-world gener- alization. In9th Annual Conference on Robot Learning

  40. [40]

    H. Wu, Y . Jing, C. Cheang, G. Chen, J. Xu, X. Li, M. Liu, H. Li, and T. Kong. Unleashing large- scale video generative pre-training for visual robot manipulation. InInternational Conference on Learning Representations, volume 2024, pages 10641–10662, 2024

    2024

  41. [41]

    K. Zhou, Y . Chen, F. Zhan, H. Hua, G. Chen, X. Chang, A. Qu, Y . Du, Z. Liu, P. P. Liang, and M. Wang. Gem-4d: Geometry-enhanced video world models for robot manipulation.arXiv preprint arXiv:2605.22882, 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  42. [42]

    A. Zeng, P. Florence, J. Tompson, S. Welker, J. Chien, M. Attarian, T. Armstrong, I. Krasin, D. Duong, V . Sindhwani, et al. Transporter networks: Rearranging the visual world for robotic manipulation. InConference on Robot Learning, pages 726–747. PMLR, 2021

    2021

  43. [43]

    Shridhar, L

    M. Shridhar, L. Manuelli, and D. Fox. Cliport: What and where pathways for robotic manipu- lation. InConference on robot learning, pages 894–906. PMLR, 2022

    2022

  44. [44]

    K. Mo, L. J. Guibas, M. Mukadam, A. Gupta, and S. Tulsiani. Where2act: From pixels to actions for articulated 3d objects. InProceedings of the IEEE/CVF International Conference on Computer Vision, pages 6813–6823, 2021

    2021

  45. [45]

    Shridhar, L

    M. Shridhar, L. Manuelli, and D. Fox. Perceiver-actor: A multi-task transformer for robotic manipulation. InConference on Robot Learning, pages 785–799. PMLR, 2023

    2023

  46. [46]

    Goyal, J

    A. Goyal, J. Xu, Y . Guo, V . Blukis, Y .-W. Chao, and D. Fox. Rvt: Robotic view transformer for 3d object manipulation. InConference on Robot Learning, pages 694–710. PMLR, 2023

    2023

  47. [47]

    Gervet, Z

    T. Gervet, Z. Xian, N. Gkanatsios, and K. Fragkiadaki. Act3d: 3d feature field transformers for multi-task robotic manipulation. In J. Tan, M. Toussaint, and K. Darvish, editors,Proceedings of The 7th Conference on Robot Learning, volume 229 ofProceedings of Machine Learning Research, pages 3949–3965. PMLR, 06–09 Nov 2023. URLhttps://proceedings.mlr. pres...

    2023

  48. [48]

    Huang, C

    W. Huang, C. Wang, R. Zhang, Y . Li, J. Wu, and L. Fei-Fei. V oxposer: Composable 3d value maps for robotic manipulation with language models. InConference on Robot Learning, pages 540–562. PMLR, 2023

    2023

  49. [49]

    Embodied Interpretability: Linking Causal Understanding to Generalization in Vision-Language-Action Models

    H. Zhang, M. Xu, A. Dhafer, S. Yue, H. Dong, and Z. D. Hao. Embodied interpretability: Link- ing causal understanding to generalization in vision-language-action models.arXiv preprint arXiv:2605.00321, 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  50. [50]

    H. R. Walke, K. Black, T. Z. Zhao, Q. Vuong, C. Zheng, P. Hansen-Estruch, A. W. He, V . My- ers, M. J. Kim, M. Du, et al. Bridgedata v2: A dataset for robot learning at scale. In7th Annual Conference on Robot Learning

  51. [51]

    Zhang, Z

    S. Zhang, Z. Xu, P. Liu, X. Yu, Y . Li, Q. Gao, Z. Fei, Z. Yin, Z. Wu, Y .-G. Jiang, et al. Vlabench: A large-scale benchmark for language-conditioned robotics manipulation with long-horizon reasoning tasks, 2024a.URL https://arxiv. org/abs/2412.18194

    work page arXiv

  52. [52]

    T. Chen, Z. Chen, B. Chen, Z. Cai, Y . Liu, Z. Li, Q. Liang, X. Lin, Y . Ge, Z. Gu, et al. Robotwin 2.0: A scalable data generator and benchmark with strong domain randomization for robust bimanual robotic manipulation.arXiv preprint arXiv:2506.18088, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  53. [53]

    S. Fei, S. Wang, J. Shi, Z. Dai, J. Cai, P. Qian, L. Ji, X. He, S. Zhang, Z. Fei, et al. Libero-plus: In-depth robustness analysis of vision-language-action models.arXiv preprint arXiv:2510.13626, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  54. [54]

    B. Liu, Y . Zhu, C. Gao, Y . Feng, Q. Liu, Y . Zhu, and P. Stone. Libero: Benchmarking knowl- edge transfer for lifelong robot learning.Advances in Neural Information Processing Systems, 36:44776–44791, 2023

    2023

  55. [55]

    FAST: Efficient Action Tokenization for Vision-Language-Action Models

    K. Pertsch, K. Stachowicz, B. Ichter, D. Driess, S. Nair, Q. Vuong, O. Mees, C. Finn, and S. Levine. Fast: Efficient action tokenization for vision-language-action models.arXiv preprint arXiv:2501.09747, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  56. [56]

    Zhong, Y

    L. Zhong, Y . Liu, Y . Wei, Z. Xiong, M. Yao, S. Liu, and G. Ren. Acot-vla: Action chain-of- thought for vision-language-action models.The IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2026

    2026

  57. [57]

    J. Cen, C. Yu, H. Yuan, Y . Jiang, S. Huang, J. Guo, X. Li, Y . Song, H. Luo, F. Wang, et al. Worldvla: Towards autoregressive action world model.arXiv preprint arXiv:2506.21539, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  58. [58]

    C.-Y . Hung, Q. Sun, P. Hong, A. Zadeh, C. Li, U. Tan, N. Majumder, S. Poria, et al. Nora: A small open-sourced generalist vision language action model for embodied tasks.arXiv preprint arXiv:2504.19854, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  59. [59]

    Q. Bu, Y . Yang, J. Cai, S. Gao, G. Ren, M. Yao, P. Luo, and H. Li. Univla: Learning to act anywhere with task-centric latent actions.arXiv preprint arXiv:2505.06111, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  60. [60]

    T. Yuan, Z. Dong, Y . Liu, and H. Zhao. Fast-wam: Do world action models need test-time future imagination?arXiv preprint arXiv:2603.16666, 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  61. [61]

    S. Tan, K. Dou, Y . Zhao, and P. Kraehenbuehl. Interactive post-training for vision-language- action models. InWorkshop on Foundation Models Meet Embodied Agents at CVPR 2025

    2025

  62. [62]

    M. J. Kim, C. Finn, and P. Liang. Fine-tuning vision-language-action models: Optimizing speed and success.arXiv preprint arXiv:2502.19645, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  63. [63]

    Y . Ze, G. Zhang, K. Zhang, C. Hu, M. Wang, and H. Xu. 3d diffusion policy: Generalizable visuomotor policy learning via simple 3d representations.Robotics: Science and Systems, 2024. 13

    2024

  64. [64]

    S. Liu, L. Wu, B. Li, H. Tan, H. Chen, Z. Wang, K. Xu, H. Su, and J. Zhu. Rdt-1b: a diffu- sion foundation model for bimanual manipulation. InInternational Conference on Learning Representations, volume 2025, pages 29982–30009, 2025

    2025

  65. [65]

    Y . Feng, H. Tan, X. Mao, C. Xiang, G. Liu, S. Huang, H. Su, and J. Zhu. Vidar: Embodied video diffusion model for generalist manipulation.arXiv preprint arXiv:2507.12898, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  66. [66]

    Y . Gui, Y . Zhou, S. Cheng, X. Yuan, H. Fan, P. Cheng, and S. Liu. Seedpolicy: Horizon scaling via self-evolving diffusion policy for robot manipulation.arXiv preprint arXiv:2603.05117, 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  67. [67]

    J. Sun, W. Zhang, Z. Qi, S. Ren, Z. Liu, H. Zhu, G. Sun, X. Jin, and Z. Chen. Vla-jepa: Enhanc- ing vision-language-action model with latent world model.arXiv preprint arXiv:2602.10098, 2026. 14 Appendix A Real-Robot Experimental Setup, Task Definitions, and Evaluation Protocol Side Camera Front Side Camera Wrist Camera Left Arm Right Arm Fig5 D405 D435i ...

    work page arXiv 2026