pith. sign in

arxiv: 2606.10568 · v1 · pith:XO65ZB6Wnew · submitted 2026-06-09 · 💻 cs.RO

VeriSpace: Spatially Grounded Action Verification for Vision-Language-Action Models

Guiyu Zhao , Longteng Guo , Junyou Zhu , Jun Fu , Yanghong Mei , Bin Cao , Jie Jiang , Xingjian He
show 1 more author
Jing Liu
This is my paper

Pith reviewed 2026-06-27 12:46 UTC · model grok-4.3

classification 💻 cs.RO
keywords vision-language-action modelsaction verification3D scene encodingrobotic manipulationtest-time verificationspatial reasoningVLA policies
0
0 comments X

The pith

VeriSpace adds dual-path 3D scene encoding and spatial reasoning to verify candidate actions before execution in VLA robotic systems.

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

The paper presents VeriSpace as a test-time verifier that lets VLA models propose multiple actions and then pick the best one instead of committing to a single prediction. It builds a scene representation that keeps both visual features and explicit 3D geometry, then reasons over spatial relations, geometric validity, and whether the action advances the task goal. This matters for robotic manipulation because small action errors often cause grasp failures or collisions, and the verifier works with any existing VLA policy without retraining. Experiments on benchmarks and real robots show higher success rates in both familiar and shifted conditions compared with base models and earlier verification approaches.

Core claim

VeriSpace evaluates candidate actions for VLA systems through Dual-Path 3D-Injected Scene Encoding, which constructs a joint visual-semantic and explicit 3D geometric scene representation, and Spatially-Grounded Action Reasoning, which assesses each action on task-relevant spatial relations, geometric validity, and expected goal progress, enabling reliable selection among subtle yet outcome-critical candidates while remaining compatible with existing VLA policies.

What carries the argument

Dual-Path 3D-Injected Scene Encoding paired with Spatially-Grounded Action Reasoning, which together evaluate actions by preserving 3D geometry and checking progress toward the task goal.

If this is right

  • Decision reliability increases over both the underlying VLA policy and earlier verification methods.
  • Performance gains appear in both in-distribution and out-of-distribution robotic tasks.
  • The verifier integrates with existing VLA policies without requiring retraining or architectural changes.
  • The same components apply to public benchmarks and real-world manipulation setups.

Where Pith is reading between the lines

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

  • Test-time verification of this form could reduce reliance on perfect one-shot prediction from the base model.
  • The modular design suggests the verifier could be stacked on top of future VLA improvements.
  • In deployed systems the approach might catch physically unsafe actions before they reach the robot.

Load-bearing premise

The dual-path encoding and spatial reasoning modules can reliably detect small geometric differences between candidate actions and judge whether an action advances the overall task goal.

What would settle it

A controlled test set of action candidates that differ only in subtle 3D geometry where VeriSpace selects the wrong action at rates no better than random or the base VLA policy.

Figures

Figures reproduced from arXiv: 2606.10568 by Bin Cao, Guiyu Zhao, Jie Jiang, Jing Liu, Jun Fu, Junyou Zhu, Longteng Guo, Xingjian He, Yanghong Mei.

Figure 1
Figure 1. Figure 1: Towards Reliable Test-Time Action Verification. (A) Instead of directly executing a single predicted action, VeriSpace [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Overview of VeriSpace. At test time, a frozen VLA policy first samples multiple candidate actions from the current [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Spatially-grounded action reasoning process. [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Real-world setup and qualitative results. Left: the robotic platform, including a Franka arm equipped with wrist [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Success rate and inference runtime as the number [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Our pipeline for generating synthetic embodied [PITH_FULL_IMAGE:figures/full_fig_p009_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Visualization results of depth estimation. [PITH_FULL_IMAGE:figures/full_fig_p010_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Visualization results of task-relevant objects grounding. [PITH_FULL_IMAGE:figures/full_fig_p010_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Examples of CoT data annotation. USER: <img> shows the current observation from the robot’s third-person camera. The robot manipulation arm is attempting to <instruction>. What action should the robot take to effectively accomplish the task? ASSISTANT: The robot may take the action <action_holder> in the next step. USER: Please evaluate the quality of the robot action. Before evaluating the action, explici… view at source ↗
Figure 10
Figure 10. Figure 10: Prompt template for candidate action scoring. [PITH_FULL_IMAGE:figures/full_fig_p012_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: The scene setting of the real-world experiments. [PITH_FULL_IMAGE:figures/full_fig_p012_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Qualitative results of our VeriSpace on real-world tasks. Representative examples include [PITH_FULL_IMAGE:figures/full_fig_p016_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: Qualitative results of our VeriSpace on the SimplerEnv-WidowX benchmark [ [PITH_FULL_IMAGE:figures/full_fig_p017_13.png] view at source ↗
read the original abstract

Vision-language-action (VLA) models have shown strong promise for robotic manipulation, but their reliability at test time remains limited by one-shot action prediction, where even small action errors can cause grasp failure, collision, or incorrect task progression. A natural alternative is to equip VLA systems with test-time verification, allowing multiple candidate actions to be proposed and evaluated before execution. However, reliable action verification is challenging because it requires not only distinguishing subtle geometric differences between candidate actions, but also assessing whether an action makes meaningful progress toward the task goal. We present VeriSpace, a 3D-aware action verifier for test-time action selection in VLA systems. VeriSpace evaluates candidate actions through two key components: Dual-Path 3D-Injected Scene Encoding, which constructs a scene representation that jointly preserves visual semantics and explicit 3D geometry, and Spatially-Grounded Action Reasoning, which evaluates each action by reasoning over task-relevant spatial relations, geometric validity, and expected goal progress. Together, these components enable more reliable discrimination between subtle yet outcome-critical action candidates while remaining fully compatible with existing VLA policies. Experiments on public benchmarks and real-world robotic manipulation tasks show that VeriSpace consistently improves decision reliability over both underlying VLA policies and prior verification-based methods, yielding substantial gains in both in-distribution and out-of-distribution settings.

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

1 major / 0 minor

Summary. The manuscript introduces VeriSpace, a test-time action verifier for vision-language-action (VLA) models in robotic manipulation. It consists of two components: Dual-Path 3D-Injected Scene Encoding, which builds a joint visual-semantic and explicit 3D geometric scene representation, and Spatially-Grounded Action Reasoning, which scores candidate actions on spatial relations, geometric validity, and task-goal progress. The verifier is designed to be compatible with existing VLA policies by evaluating multiple proposed actions before execution. The central claim is that this yields consistent reliability gains over base VLA policies and prior verification methods on public benchmarks and real-world tasks, in both in-distribution and out-of-distribution regimes.

Significance. If the claimed improvements are supported by rigorous experiments, the work would address a practical limitation of one-shot VLA prediction by adding lightweight, 3D-aware verification. The emphasis on compatibility with unmodified VLA policies and the focus on subtle geometric distinctions are plausible contributions to reliable robotic control. However, the provided manuscript text contains only the abstract and no methods, equations, ablations, quantitative tables, or experimental protocols, preventing any assessment of whether the dual-path encoding or spatial reasoning components actually deliver the stated gains.

major comments (1)
  1. [Abstract] Abstract: The central empirical claim (consistent improvements over VLA baselines and prior verifiers in ID/OOD settings) cannot be evaluated because the manuscript supplies no experimental setup, baselines, metrics, or results. This is load-bearing for the paper's contribution.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their review. We address the single major comment below, noting that the full manuscript (beyond the abstract) contains the requested details.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The central empirical claim (consistent improvements over VLA baselines and prior verifiers in ID/OOD settings) cannot be evaluated because the manuscript supplies no experimental setup, baselines, metrics, or results. This is load-bearing for the paper's contribution.

    Authors: The full manuscript includes dedicated Methods sections describing Dual-Path 3D-Injected Scene Encoding (with explicit 3D geometry injection and joint visual-semantic representations) and Spatially-Grounded Action Reasoning (with scoring over spatial relations, geometric validity, and goal progress). The Experiments section specifies public benchmarks, real-world tasks, baselines (base VLA policies and prior verification methods), metrics (success rate, reliability gains), and quantitative results demonstrating consistent improvements in both in-distribution and out-of-distribution regimes. These sections were available in the complete submission; if only the abstract was reviewed, we can supply the relevant excerpts or page references. revision: no

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The provided abstract and context describe VeriSpace as an additive test-time verifier with two new components (Dual-Path 3D-Injected Scene Encoding and Spatially-Grounded Action Reasoning) layered on top of existing VLA policies. No derivation chain, equations, fitted parameters presented as predictions, uniqueness theorems, or self-citations are present. The central claim rests on experimental gains rather than any self-referential construction or reduction to inputs by definition. This is a standard non-circular presentation of an engineering addition to prior models.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Only the abstract is available; no information on free parameters, axioms, or invented entities is provided.

pith-pipeline@v0.9.1-grok · 5791 in / 1085 out tokens · 19989 ms · 2026-06-27T12:46:03.413684+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

51 extracted references · 19 linked inside Pith

  1. [1]

    Suneel Belkhale, Tianli Ding, Ted Xiao, Pierre Sermanet, Quon Vuong, Jonathan Tompson, Yevgen Chebotar, Debidatta Dwibedi, and Dorsa Sadigh. 2024. Rt-h: Action hierarchies using language.arXiv preprint arXiv:2403.01823(2024)

    Pith/arXiv arXiv 2024

  2. [2]

    Lucas Beyer, Andreas Steiner, André Susano Pinto, Alexander Kolesnikov, Xiao Wang, Daniel Salz, Maxim Neumann, Ibrahim Alabdulmohsin, Michael Tschan- nen, Emanuele Bugliarello, et al. 2024. Paligemma: A versatile 3b vlm for transfer. arXiv preprint arXiv:2407.07726(2024)

    Pith/arXiv arXiv 2024

  3. [3]

    Kevin Black, Noah Brown, Danny Driess, Adnan Esmail, Michael Equi, Chelsea Finn, Niccolo Fusai, Lachy Groom, Karol Hausman, Brian Ichter, et al. 2024. pi0: A Vision-Language-Action Flow Model for General Robot Control.arXiv preprint arXiv:2410.24164(2024)

    Pith/arXiv arXiv 2024

  4. [4]

    Anthony Brohan, Noah Brown, Justice Carbajal, Yevgen Chebotar, Joseph Dabis, Chelsea Finn, Keerthana Gopalakrishnan, Karol Hausman, Alex Herzog, Jasmine Hsu, et al. 2022. Rt-1: Robotics transformer for real-world control at scale.arXiv preprint arXiv:2212.06817(2022)

    Pith/arXiv arXiv 2022

  5. [5]

    Bradley Brown, Jordan Juravsky, Ryan Ehrlich, Ronald Clark, Quoc V Le, Christo- pher Ré, and Azalia Mirhoseini. 2024. Large language monkeys: Scaling inference compute with repeated sampling.arXiv preprint arXiv:2407.21787(2024)

    Pith/arXiv arXiv 2024

  6. [6]

    Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, et al. 2020. Language models are few-shot learners.NeurIPS33 (2020), 1877–1901

    2020

  7. [7]

    Guoxin Chen, Minpeng Liao, Chengxi Li, and Kai Fan. 2024. Alphamath almost zero: Process supervision without process, 2024.URL https://arxiv. org/abs/2405.03553(2024)

    arXiv 2024

  8. [8]

    An-Chieh Cheng, Hongxu Yin, Yang Fu, Qiushan Guo, Ruihan Yang, Jan Kautz, Xiaolong Wang, and Sifei Liu. 2024. Spatialrgpt: Grounded spatial reasoning in vision-language models.Advances in Neural Information Processing Systems37 (2024), 135062–135093

    2024

  9. [9]

    Mingtong Dai, Lingbo Liu, Yongjie Bai, Yang Liu, Zhouxia Wang, Rui Su, Chunjie Chen, Liang Lin, and Xinyu Wu. 2025. RoVer: Robot Reward Model as Test- Time Verifier for Vision-Language-Action Model.arXiv preprint arXiv:2510.10975 (2025)

    arXiv 2025

  10. [10]

    Runpei Dong, Chunrui Han, Yuang Peng, Zekun Qi, Zheng Ge, Jinrong Yang, Liang Zhao, Jianjian Sun, Hongyu Zhou, Haoran Wei, Xiangwen Kong, Xiangyu Zhang, Kaisheng Ma, and Li Yi. 2024. DreamLLM: Synergistic Multimodal Com- prehension and Creation. InICLR

    2024

  11. [11]

    David Heineman, Yao Dou, and Wei Xu. 2024. Improving minimum Bayes risk decoding with multi-prompt. InProceedings of the 2024 Conference on Empirical Methods in Natural Language Processing. 22525–22545

    2024

  12. [12]

    Edward J Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Liang Wang, Weizhu Chen, et al. 2022. Lora: Low-rank adaptation of large language models.Iclr1, 2 (2022), 3

    2022

  13. [13]

    Physical Intelligence, Kevin Black, Noah Brown, James Darpinian, Karan Dhabalia, Danny Driess, Adnan Esmail, Michael Equi, Chelsea Finn, Niccolo Fusai, et al

  14. [14]

    arXiv preprint arXiv:2504.16054(2025)

    pi0.5: a Vision-Language-Action Model with Open-World Generalization. arXiv preprint arXiv:2504.16054(2025)

    Pith/arXiv arXiv 2025

  15. [15]

    Suhyeok Jang, Dongyoung Kim, Changyeon Kim, Youngsuk Kim, and Jinwoo Shin. 2025. Verifier-free Test-Time Sampling for Vision Language Action Models. arXiv preprint arXiv:2510.05681(2025)

    arXiv 2025

  16. [16]

    Alexander Khazatsky, Karl Pertsch, Suraj Nair, Ashwin Balakrishna, Sudeep Dasari, Siddharth Karamcheti, Soroush Nasiriany, Mohan Kumar Srirama, Lawrence Yunliang Chen, Kirsty Ellis, et al. 2024. Droid: A large-scale in-the-wild robot manipulation dataset.arXiv preprint arXiv:2403.12945(2024)

    Pith/arXiv arXiv 2024

  17. [17]

    Moo Jin Kim, Chelsea Finn, and Percy Liang. 2025. Fine-tuning vision-language- action models: Optimizing speed and success.arXiv preprint arXiv:2502.19645 (2025)

    Pith/arXiv arXiv 2025

  18. [18]

    Moo Jin Kim, Karl Pertsch, Siddharth Karamcheti, Ted Xiao, Ashwin Balakr- ishna, Suraj Nair, Rafael Rafailov, Ethan Foster, Grace Lam, Pannag Sanketi, et al

  19. [19]

    Openvla: An open-source vision-language-action model.arXiv preprint arXiv:2406.09246(2024)

    Pith/arXiv arXiv 2024

  20. [20]

    Jacky Kwok, Christopher Agia, Rohan Sinha, Matt Foutter, Shulu Li, Ion Stoica, Azalia Mirhoseini, and Marco Pavone. 2025. Robomonkey: Scaling test-time sampling and verification for vision-language-action models.arXiv preprint arXiv:2506.17811(2025)

    arXiv 2025

  21. [21]

    Tony Lee, Andrew Wagenmaker, Karl Pertsch, Percy Liang, Sergey Levine, and Chelsea Finn. 2026. RoboReward: General-Purpose Vision-Language Reward Models for Robotics.arXiv preprint arXiv:2601.00675(2026)

    arXiv 2026

  22. [22]

    Chengmeng Li, Junjie Wen, Yaxin Peng, Yan Peng, and Yichen Zhu. 2026. Pointvla: Injecting the 3d world into vision-language-action models.IEEE Robotics and Automation Letters11, 3 (2026), 2506–2513

    2026

  23. [23]

    Peiyan Li, Yixiang Chen, Hongtao Wu, Xiao Ma, Xiangnan Wu, Yan Huang, Liang Wang, Tao Kong, and Tieniu Tan. 2025. Bridgevla: Input-output alignment for efficient 3d manipulation learning with vision-language models.arXiv preprint arXiv:2506.07961(2025). VeriSpace: Spatially Grounded Action Verification for Vision-Language-Action Models Conference acronym ...

    arXiv 2025

  24. [24]

    Xiaoqi Li, Liang Heng, Jiaming Liu, Yan Shen, Chenyang Gu, Zhuoyang Liu, Hao Chen, Nuowei Han, Renrui Zhang, Hao Tang, et al . 2025. 3ds-vla: A 3d spatial-aware vision language action model for robust multi-task manipulation. In9th Annual Conference on Robot Learning

    2025

  25. [25]

    Xuanlin Li, Kyle Hsu, Jiayuan Gu, Karl Pertsch, Oier Mees, Homer Rich Walke, Chuyuan Fu, Ishikaa Lunawat, Isabel Sieh, Sean Kirmani, Sergey Levine, Jiajun Wu, Chelsea Finn, Hao Su, Quan Vuong, and Ted Xiao. 2024. Evaluating Real- World Robot Manipulation Policies in Simulation.arXiv preprint arXiv:2405.05941 (2024)

    Pith/arXiv arXiv 2024

  26. [26]

    Xinghang Li, Peiyan Li, Minghuan Liu, Dong Wang, Jirong Liu, Bingyi Kang, Xiao Ma, Tao Kong, Hanbo Zhang, and Huaping Liu. 2024. Towards generalist robot policies: What matters in building vision-language-action models.arXiv preprint arXiv:2412.14058(2024)

    Pith/arXiv arXiv 2024

  27. [27]

    Tao Lin, Gen Li, Yilei Zhong, Yanwen Zou, Yuxin Du, Jiting Liu, Encheng Gu, and Bo Zhao. 2025. Evo-0: Vision-language-action model with implicit spatial understanding.arXiv preprint arXiv:2507.00416(2025)

    arXiv 2025

  28. [28]

    Bo Liu, Yifeng Zhu, Chongkai Gao, Yihao Feng, Qiang Liu, Yuke Zhu, and Peter Stone. 2023. Libero: Benchmarking knowledge transfer for lifelong robot learning. Advances in Neural Information Processing Systems36 (2023), 44776–44791

    2023

  29. [29]

    Haotian Liu, Chunyuan Li, Qingyang Wu, and Yong Jae Lee. 2023. Visual in- struction tuning.Advances in neural information processing systems36 (2023), 34892–34916

    2023

  30. [30]

    Jiaming Liu, Hao Chen, Pengju An, Zhuoyang Liu, Renrui Zhang, Chenyang Gu, Xiaoqi Li, Ziyu Guo, Sixiang Chen, Mengzhen Liu, et al . 2025. Hybridvla: Collaborative diffusion and autoregression in a unified vision-language-action model.arXiv preprint arXiv:2503.10631(2025)

    Pith/arXiv arXiv 2025

  31. [31]

    Songming Liu, Lingxuan Wu, Bangguo Li, Hengkai Tan, Huayu Chen, Zhengyi Wang, Ke Xu, Hang Su, and Jun Zhu. 2024. Rdt-1b: a diffusion foundation model for bimanual manipulation.arXiv preprint arXiv:2410.07864(2024)

    Pith/arXiv arXiv 2024

  32. [32]

    Shilong Liu, Zhaoyang Zeng, Tianhe Ren, Feng Li, Hao Zhang, Jie Yang, Qing Jiang, Chunyuan Li, Jianwei Yang, Hang Su, et al. 2024. Grounding dino: Marry- ing dino with grounded pre-training for open-set object detection. InEuropean conference on computer vision. Springer, 38–55

    2024

  33. [33]

    Mitsuhiko Nakamoto, Oier Mees, Aviral Kumar, and Sergey Levine. 2024. Steering your generalists: Improving robotic foundation models via value guidance.arXiv preprint arXiv:2410.13816(2024)

    arXiv 2024

  34. [34]

    OpenAI. 2023. GPT-4V(ision) System Card. https://openai.com/research/gpt-4v- system-card

    2023

  35. [35]

    Long Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida, Carroll Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, et al. 2022. Training language models to follow instructions with human feedback.Advances in neural information processing systems35 (2022), 27730–27744

    2022

  36. [36]

    Charles Ruizhongtai Qi, Li Yi, Hao Su, and Leonidas J Guibas. 2017. Pointnet++: Deep hierarchical feature learning on point sets in a metric space.Advances in neural information processing systems30 (2017)

    2017

  37. [37]

    Zekun Qi, Runpei Dong, Shaochen Zhang, Haoran Geng, Chunrui Han, Zheng Ge, Li Yi, and Kaisheng Ma. 2024. ShapeLLM: Universal 3D Object Understanding for Embodied Interaction. InComputer Vision - ECCV 2024 - 18th European Conference, Milan, Italy, September 29-October 4, 2024, Proceedings, Part XLIII (Lecture Notes in Computer Science, Vol. 15101). Springe...

    2024

  38. [38]

    Delin Qu, Haoming Song, Qizhi Chen, Yuanqi Yao, Xinyi Ye, Yan Ding, Zhigang Wang, JiaYuan Gu, Bin Zhao, Dong Wang, et al. 2025. Spatialvla: Exploring spatial representations for visual-language-action model.arXiv preprint arXiv:2501.15830 (2025)

    Pith/arXiv arXiv 2025

  39. [39]

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al. 2021. Learning transferable visual models from natural language supervision. InInternational conference on machine learning. PmLR, 8748–8763

    2021

  40. [40]

    Lin Sun, Bin Xie, Yingfei Liu, Hao Shi, Tiancai Wang, and Jiale Cao. 2025. Geovla: Empowering 3d representations in vision-language-action models.arXiv preprint arXiv:2508.09071(2025)

    arXiv 2025

  41. [41]

    Octo Model Team, Dibya Ghosh, Homer Walke, Karl Pertsch, Kevin Black, Oier Mees, Sudeep Dasari, Joey Hejna, Tobias Kreiman, Charles Xu, et al. 2024. Octo: An open-source generalist robot policy.arXiv preprint arXiv:2405.12213(2024)

    Pith/arXiv arXiv 2024

  42. [42]

    Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, Marie-Anne Lachaux, Timothée Lacroix, Baptiste Rozière, Naman Goyal, Eric Hambro, Faisal Azhar, et al. 2023. Llama: Open and efficient foundation language models. arXiv 2023.arXiv preprint arXiv:2302.1397110 (2023)

    Pith/arXiv arXiv 2023

  43. [43]

    Homer Rich Walke, Kevin Black, Tony Z Zhao, Quan Vuong, Chongyi Zheng, Philippe Hansen-Estruch, Andre Wang He, Vivek Myers, Moo Jin Kim, Max Du, et al. 2023. Bridgedata v2: A dataset for robot learning at scale. InConference on Robot Learning. PMLR, 1723–1736

    2023

  44. [44]

    Xuezhi Wang, Jason Wei, Dale Schuurmans, Quoc Le, Ed Chi, Sharan Narang, Aakanksha Chowdhery, and Denny Zhou. 2022. Self-consistency improves chain of thought reasoning in language models.arXiv preprint arXiv:2203.11171(2022)

    Pith/arXiv arXiv 2022

  45. [45]

    Junjie Wen, Yichen Zhu, Jinming Li, Minjie Zhu, Zhibin Tang, Kun Wu, Zhiyuan Xu, Ning Liu, Ran Cheng, Chaomin Shen, et al. 2025. Tinyvla: Towards fast, data- efficient vision-language-action models for robotic manipulation.IEEE Robotics and Automation Letters(2025)

    2025

  46. [46]

    Kun Wu, Chengkai Hou, Jiaming Liu, Zhengping Che, Xiaozhu Ju, Zhuqin Yang, Meng Li, Yinuo Zhao, Zhiyuan Xu, Guang Yang, et al. 2024. Robomind: Benchmark on multi-embodiment intelligence normative data for robot manipulation.arXiv preprint arXiv:2412.13877(2024)

    Pith/arXiv arXiv 2024

  47. [47]

    Lihe Yang, Bingyi Kang, Zilong Huang, Zhen Zhao, Xiaogang Xu, Jiashi Feng, and Hengshuang Zhao. 2024. Depth anything v2.Advances in Neural Information Processing Systems37 (2024), 21875–21911

    2024

  48. [48]

    Jiahui Zhang, Yurui Chen, Yueming Xu, Ze Huang, Yanpeng Zhou, Yu-Jie Yuan, Xinyue Cai, Guowei Huang, Xingyue Quan, Hang Xu, et al. 2025. 4d-vla: Spa- tiotemporal vision-language-action pretraining with cross-scene calibration. arXiv preprint arXiv:2506.22242(2025)

    arXiv 2025

  49. [49]

    Zhengshen Zhang, Hao Li, Yalun Dai, Zhengbang Zhu, Lei Zhou, Chenchen Liu, Dong Wang, Francis EH Tay, Sijin Chen, Ziwei Liu, et al. 2025. From spatial to actions: Grounding vision-language-action model in spatial foundation priors. arXiv preprint arXiv:2510.17439(2025)

    arXiv 2025

  50. [50]

    Haoyu Zhen, Xiaowen Qiu, Peihao Chen, Jincheng Yang, Xin Yan, Yilun Du, Yin- ing Hong, and Chuang Gan. 2024. 3d-vla: A 3d vision-language-action generative world model.arXiv preprint arXiv:2403.09631(2024)

    Pith/arXiv arXiv 2024

  51. [51]

    Brianna Zitkovich, Tianhe Yu, Sichun Xu, Peng Xu, Ted Xiao, Fei Xia, Jialin Wu, Paul Wohlhart, Stefan Welker, Ayzaan Wahid, et al. 2023. Rt-2: Vision-language- action models transfer web knowledge to robotic control. InConference on Robot Learning. PMLR, 2165–2183. Conference acronym ’XX, June 03–05, 2018, Woodstock, NY Guiyu Zhao1,2, Longteng Guo1, Junyo...

    2023