Pith. sign in

REVIEW 4 major objections 5 minor 35 references

RynnValue: Scaling Robotic Value Foundation Models with Temporal Distance

T0 review · 4 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read RynnValue replaces preference annotations with timestamp-derived temporal distance, and the resulting value model beats fully preference-supervised rivals on out-of-distribution trajectory ranking.

desk verdict A serious scaling recipe for temporal-distance reward, but the headline OOD claim lacks a documented train/eval separation and the real-world gains are confounded by per-method shaping coefficients. read the letter →

arxiv 2608.09853 v1 pith:I3CC6RMH submitted 2026-08-10 cs.RO cs.CVcs.LG

Dongchi Huang , Hongyin Zhang , Bohan Hou , Siteng Huang , Zhian Su , Hang Guo , Tong Lu , Zhaofeng Xu
show 7 more authors
Jiahao Tang Jianfei Yang Donglin Wang Peixi Peng Mingxiu Chen Deli Zhao Xin Li
This is my paper · ORCID
classification cs.ROcs.CVcs.LG
keywords temporaldistancevaluefoundationmodelrobotmanipulationrewardpotential-basedshapingpreference-freelearningzero-shotgeneralizationtrajectoryranking
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

General-purpose reward models are the bottleneck for scaling robot learning, but their usual supervision targets, preferences and normalized progress, tie a value model to trajectory-internal anchors that do not transfer across embodiments and data sources. RynnValue replaces those anchors with temporal distance, the directed cost-to-go from an observation to the language-specified goal, and shows that such labels can be read directly from timestamps, needing neither preference pairs nor progress annotations. Trained on over 7,000 hours and roughly 3M instruction-conditioned clips, the resulting value model attains an average Kendall's $\tau_a$ of 0.675 on the RBM-EVAL-OOD trajectory-ranking suite, surpassing the 0.655 of the fully preference-supervised state of the art while generalizing zero-shot to unseen tasks, embodiments, and viewpoints. Converted into dense rewards through potential-based shaping, it raises real-world policy success from 52.5% to 72.5% online and from 63.8% to 82.5% offline. If these results hold, temporal distance is a more scalable supervision target and a practical reward interface for generalist robot policies.

What carries the argument

The central object is temporal distance, defined as the directed cost-to-go from an observation to the language-specified goal, with labels derived directly from timestamps once a semantic completion cutoff is identified; it carries the argument because it replaces preferences and normalized progress with a single transferable supervision target. The supporting mechanism in the model is value-isolation attention, which prevents each group of repeated value-query tokens from attending to other groups' value predictions, together with random temporal sampling and temporal-order shuffling that break the correspondence between sequence position and task progress. At inference, predicted remaining time is negated into a potential, and potential-based shaping turns it into a dense reward for policy learning.

What would settle it

Match every episode in the six RBM-EVAL-OOD datasets against the training corpora listed in Table 1, for example by source dataset name, episode hash, or URL. If even a fraction of evaluation episodes occur in training, the zero-shot out-of-distribution and state-of-the-art claims would need to be recomputed, and a positive control would be to retrain on a corpus with those episodes explicitly held out to see whether the 0.675 average $\tau_a$ persists.

Watch

Extended reading notes

Core claim

The paper's central claim is that goal-conditioned temporal distance, rather than preferences or normalized progress, is the right scaling target for robotic value models. Trained without any trajectory-level preference labels, RynnValue predicts the remaining time to a relabeled completion cutoff along with signed temporal displacement between presented observations, using distributional heads over symlog bins and a set of shortcut-suppression designs: random temporal sampling, temporal-order shuffling, and value-isolation attention. The authors report that this recipe, not model scale, carries the gains: a 4B and an 8B variant perform comparably, and ablations show that removing shuffling, isolation, random sampling, or language supervision each degrades the average ranking score. The paper concludes that temporal distance unifies heterogeneous data under a single preference-free value interface and that it serves as a practical dense reward source for downstream online and offline reinforcement learning.

Load-bearing premise

The claim depends on the assumption that none of the six RBM-EVAL-OOD evaluation datasets appear in the training mixture; the paper asserts they are unseen but does not document an exclusion procedure for them.

Editorial extensions

If this is right

  • Temporal-distance labels can be produced automatically from timestamps, so training a general-purpose value model no longer requires constructing preference pairs or normalizing every task onto a $[0,1]$ progress scale.
  • A preference-free value model can outperform preference-supervised models on out-of-distribution trajectory ranking, suggesting that ordering information can be extracted from timestamp structure alone.
  • The same model serves as a dense reward function for both online and offline RL, improving real-world policy success without any target-domain fine-tuning of the reward model.
  • The scaling analysis indicates that broadening task coverage improves temporal-value generalization more than adding episodes within known tasks, pointing data collection toward task diversity.
  • The auxiliary language supervision, including video description, instruction-matching, and success prediction, contributes semantic grounding beyond the continuous temporal-value targets.

Reading between the lines

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

  • Editorial extension: the timestamp-derived temporal-distance target could plausibly be applied to human egocentric video as well as robot data, potentially yielding reward models that judge demonstration quality without preference annotations.
  • Editorial extension: because the relative temporal head is instruction-independent, it might be repurposed as a general temporal anomaly or regression detector during policy rollouts, even when the language goal is ambiguous.
  • Editorial extension: the paper's diversity-versus-volume result suggests a testable prediction that a smaller but more task-diverse corpus can beat a larger repetitive one on unseen tasks, which could guide future data-collection budgets for value models.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper introduces RynnValue, a vision-language value foundation model that predicts temporal distance (remaining time to a language-specified goal) as a replacement for preference-based or normalized-progress reward supervision. Labels are derived from timestamps after subtask segmentation and cutoff relabeling, allowing training on over 7,000 hours and roughly 3M instruction-conditioned clips without preference annotations. The model combines random temporal sampling, temporal-order shuffling, and value-isolation attention to suppress shortcuts, and uses dual distributional heads for absolute and relative temporal estimates. On the RBM-EVAL-OOD trajectory-ranking benchmark, RynnValue-8B reports average Kendall's tau_a of 0.675, surpassing the preference-supervised Robometer baseline at 0.655, and real-world manipulation experiments report improved success rates when RynnValue's temporal-distance potential is converted into a dense reward via potential-based shaping.

Significance. If the central empirical claims hold, the paper makes a strong contribution: it identifies a supervision target for value/reward models that is substantially cheaper than preference labels, demonstrates scaling to a large heterogeneous corpus, and provides architectural components (temporal-order shuffling, value-isolation attention) that are plausibly useful beyond this specific instantiation. The real-world RL results, if reproducible and properly controlled, would be practically important. The paper also ships model releases and a detailed appendix on data curation and experimental protocols, which supports reproducibility. However, the headline zero-shot generalization and state-of-the-art claims rest on an unverified train/eval exclusion and on a benchmark whose quality labels may coincide with the training target; these issues require resolution before the significance of the results can be fully assessed.

major comments (4)
  1. [§4.2, Table 1] The paper does not document that the RBM-EVAL-OOD evaluation datasets (USC Franka, USC Koch, USC Trossen, USC xArm, MIT Franka, UTD SO101) are absent from the training mixture. Table 1 lists 693,037 Open X-Embodiment episodes and several large public corpora, but no exclusion procedure, dataset-ID cross-check, or trajectory-overlap audit is provided. Since the headline claim is zero-shot out-of-distribution performance and superiority over the preference-supervised state of the art, the absence of this audit is load-bearing. The authors should either provide a verifiable exclusion procedure, report overlap statistics, or temper the zero-shot and OOD claims.
  2. [§4.2, Eq. (10)] The RBM-EVAL-OOD ranking score is the negative predicted remaining time -v_end, which is the model's own trained temporal-distance output. The benchmark's ground-truth quality labels are described as failed, suboptimal, and successful executions, which are likely completion-based. If this is so, the evaluation partly validates the training target rather than an independent notion of reward quality, and the comparison against preference-supervised baselines may be less surprising than presented. The authors should clarify what the ground-truth ordering represents and provide evidence, such as a human-preference-based evaluation or an analysis of trajectories where completion status is not aligned with temporal distance, that the benchmark measures more than the training objective itself.
  3. [§4.5, Eq. (11), Table 4] The real-world RL comparison uses different shaping coefficients for the two reward models: kappa = 0.1 for RynnValue and kappa = 1.0 for Robometer. Because kappa scales the entire potential-based shaping term, the observed success-rate differences confound reward-model quality with reward scale/gain. The manuscript does not report a sensitivity analysis over kappa for either model. The authors should show that the conclusions are stable across a range of kappa values or justify the asymmetric choice with a principled selection procedure.
  4. [§4.5, Table 4] The real-world policy results are reported as success rates over only 20 trials per task, with no confidence intervals or significance tests. For example, the offline difference between RynnValue (100%) and Robometer (80%) on Bread Basket Placement corresponds to 20 vs 16 successes out of 20 trials, which is not statistically meaningful at conventional levels. The paper should report confidence intervals, exact binomial tests, or additional trials to support the claimed improvements.
minor comments (5)
  1. [Figure 6 caption] The caption contains a duplicated sentence: 'Each row shows a sequence of observations from one manipulation task' appears twice and should be removed once.
  2. [§3.1.1] The abstract and introduction state 'over 7,000 hours', but Table 1 reports episode and segmentation counts without a direct hours total. It would help to state how the 7,000-hour figure is computed from the sources.
  3. [§4.2.1] The sentence 'RynnValue-8B ranks first on USC Franka, USC Koch, and USC Trossen' is accurate, but the next sentence about RynnValue-4B on UTD SO101 could be phrased more clearly since the 8B model achieves 0.833 there and the 4B model 0.933.
  4. [§4.2.2, Figure 3] The confusion matrices are reported only as 'normalized diagonal margin' values in the text; it would be helpful to define this metric explicitly in the main text rather than only in the figure caption.
  5. [§5] The limitations paragraph mentions the short observation window and the minimum-time objective, but it does not discuss the potential issue of temporal-distance labels being unreliable when trajectories contain post-completion motions or when the cutoff relabeling is incorrect. A brief comment on this would be useful.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: temporal-distance targets are timestamp-derived and benchmarked against external quality labels; self-citations are implementation choices, not load-bearing.

full rationale

RynnValue's derivation chain is externally anchored at each step. The training target v*_i = max(0, t_G - t_i) is read directly from timestamps and relabeled cutoffs (Eq. 6), not from the evaluation metric. The RBM-EVAL-OOD benchmark supplies external ground-truth trajectory-quality labels (failed/suboptimal/successful), and the model is scored by -v_end; this is a candidate reward interface, not a fitted input, and the model can fail this test, as the ablations in Table 3 show. The potential-based shaping conversion uses the standard Ng et al. result, and the real-world RL comparison uses externally collected success labels and a fixed shared policy backbone. The only self-citations, RynnBrain [5] as the backbone and FiRE [11] for grouped query tokens, are implementation choices and do not carry the central temporal-distance claim. The paper does assert in Section 4.5 that real-world tasks and the RBM-EVAL-OOD datasets are absent from training, but it documents no overlap audit; that is a data-leakage and correctness risk, not an equation-level circularity, because leakage would concern training-data contamination rather than the prediction reducing to its inputs by construction. No quoted equation or definition equates the benchmark score with the training target, and the claimed SOTA result is not forced by the paper's own formulation.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The paper's labels are defined by timestamps and relabeled cutoffs, so the central free choices are the cutoff rules and the reward scale. No invented physical entities are introduced. The most fragile assumptions are that completion time is a valid value signal and that the evaluation data is truly unseen.

free parameters (4)
  • per-dataset cutoff trimming ratios and durations = not disclosed
    Section 3.1.2: 'dataset-specific ratio- or duration-based trimming is applied when necessary'. These rules define the completion cutoff t_G and therefore every temporal-distance label; without them, the supervision target is underspecified.
  • shaping coefficient kappa for RynnValue = 0.1
    Appendix B.5 and B.6: kappa is fixed across tasks but differs from Robometer's 1.0. It rescales the dense reward and is chosen per method, so real-world gains may partly reflect reward scaling.
  • shaping coefficient kappa for Robometer = 1.0
    Same source as above. No common normalization or sensitivity sweep is reported.
  • rewind probability and shuffle probability = 0.3 and 0.5
    Section 4.1: training hyperparameters chosen by hand; ablations show they are important, but no sweep is reported.
assumptions (4)
  • domain assumption Timestamp to a relabeled completion cutoff is a faithful proxy for task progress and value under a minimum-time objective.
    Section 3.1.2 defines labels as max(0, t_G - t_i). This equates value with remaining time, ignoring idling, non-uniform execution speed, energy, safety, and precision; the paper itself notes the minimum-time assumption as a limitation in Section 5.
  • domain assumption Subtask segmentation and cutoff relabeling identify the first semantically completed observation.
    Section 3.1.2: segmentation uses native annotations or whole episodes, and 'dataset-specific ratio- or duration-based trimming is applied when necessary'. If cutoffs are wrong, temporal-distance labels are wrong.
  • ad hoc to paper The RBM-EVAL-OOD evaluation datasets are absent from the training mixture.
    Section 4.2 calls the benchmark out-of-distribution, but Table 1 lists Open X-Embodiment (693k episodes), which historically contains USC/MIT/UTD robot datasets; no exclusion is documented. This assumption is load-bearing for the 'zero-shot OOD' claim.
  • domain assumption RynnBrain [5] provides suitable pretrained visual-language representations.
    Section 2: the model is built on RynnBrain, a self-cited unpublished backbone; the paper does not validate that the backbone is unbiased or that its training data is disjoint from the evaluation sets.

how reviews work

0 comments
Cite this review

Pith. "Pith review of RynnValue: Scaling Robotic Value Foundation Models with Temporal Distance." pith.science (2026). https://pith.science/paper/I3CC6RMH

@misc{pith2026260809853,
  author       = {Pith},
  title        = {Pith review of: RynnValue: Scaling Robotic Value Foundation Models with Temporal Distance},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/I3CC6RMH}},
  note         = {Machine review of arXiv:2608.09853}
}
read the original abstract

General-purpose reward models are increasingly the bottleneck for scaling robot learning, yet the recipe for learning value-related capabilities from large-scale heterogeneous corpora remains underexplored. Existing approaches tie supervision to task-internal anchors such as preferences or normalized progress, none of which transfer cleanly across embodiments and data sources. We introduce RynnValue, an open-source value foundation model for robotic manipulation that replaces these anchors with temporal distance, the directed cost-to-go from an observation to the language-specified goal. Because temporal-distance labels can be derived directly from timestamps, RynnValue scales to over 7,000 hours and roughly 3M instruction-conditioned clips without preference or progress annotations. To make temporal-value learning reliable at scale, we combine random temporal sampling, temporal-order shuffling, and value-isolation attention, suppressing shortcuts that would leave predictions insensitive to failures and regressions. Trained without preference labels, RynnValue attains an average Kendall's tau_a of 0.675 on RBM-EVAL-OOD, surpassing the fully preference-supervised state of the art (0.655) and more than doubling a progress-only counterpart (0.292), while generalizing zero-shot to unseen tasks, embodiments, and viewpoints. Converted into dense rewards via potential-based shaping, it raises real-world policy success from 52.5% to 72.5% online and from 63.8% to 82.5% offline. These results establish temporal distance as a scalable supervision target and practical reward interface for generalist robot policies.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

35 extracted references · 20 canonical work pages

  1. [18]

    TimeRewarder: Learning dense reward from passive videos via frame-wise temporal distance.arXiv preprint arXiv:2509.26627, 2025

    Yuyang Liu, Chuan Wen, Yihang Hu, Dinesh Jayaraman, and Yang Gao. TimeRewarder: Learning dense reward from passive videos via frame-wise temporal distance.arXiv preprint arXiv:2509.26627, 2025

    arXiv 2025

  2. [1]

    Qwen3-VL technical report.arXiv preprint arXiv:2511.21631, 2025

    Shuai Bai, Yuxuan Cai, Ruizhe Chen, Keqin Chen, Xionghui Chen, Zesen Cheng, Lianghao Deng, Wei Ding, Chang Gao, Chunjiang Ge, Wenbin Ge, Zhifang Guo, Qidong Huang, Jie Huang, Fei Huang, Binyuan Hui, Shutong Jiang, Zhaohai Li, Mingsheng Li, Mei Li, Kaixin Li, Zicheng Lin, Junyang Lin, Xuejing Liu, Jiawei Liu, Chenglong Liu, Yang Liu, Dayiheng Liu, Shixuan ...

    arXiv 2025

  3. [2]

    InProceedings of the Conference on Robot Learning, 2025

    Kevin Black, Noah Brown, James Darpinian, Karan Dhabalia, Danny Driess, Adnan Esmail, Michael Robert Equi, Chelsea Finn, Niccolo Fusai, Manuel Y Galliker, et al.π0.5: A Vision-Language-Action Model with Open-World Generalization. InProceedings of the Conference on Robot Learning, 2025

    work page 2025

  4. [3]

    AgiBot World Colosseum: A large-scale manipulation platform for scalable and intelligent embodied systems

    Qingwen Bu, Jisong Cai, Li Chen, Xiuqi Cui, Yan Ding, Siyuan Feng, Xindong He, Xu Huang, et al. AgiBot World Colosseum: A large-scale manipulation platform for scalable and intelligent embodied systems. InProceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems, 2025

    work page 2025

  5. [4]

    LeRobot: State-of-the-art machine learning for real-world robotics in PyTorch.https://github.com/huggingface/lerobot, 2024

    Remi Cadene, Simon Alibert, Alexander Soare, Quentin Gallouedec, Adil Zouitine, Steven Palma, Pepijn Kooijmans, Michel Aractingi, Mustafa Shukor, Dana Aubakirova, Martino Russi, Francesco Capuano, Caroline Pascal, Jade Choghari, Jess Moss, and Thomas Wolf. LeRobot: State-of-the-art machine learning for real-world robotics in PyTorch.https://github.com/hug...

    work page 2024

  6. [5]

    RynnBrain: Open embodied foundation models.arXiv preprint arXiv:2602.14979, 2026

    Ronghao Dang, Jiayan Guo, Bohan Hou, Sicong Leng, Kehan Li, Xin Li, Jiangpin Liu, Yunxuan Mao, Zhikai Wang, Yuqian Yuan, et al. RynnBrain: Open embodied foundation models.arXiv preprint arXiv:2602.14979, 2026

    arXiv 2026

  7. [6]

    Unified language model pre-training for natural language understanding and generation

    Li Dong, Nan Yang, Wenhui Wang, Furu Wei, Xiaodong Liu, Yu Wang, Jianfeng Gao, Ming Zhou, and Hsiao-Wuen Hon. Unified language model pre-training for natural language understanding and generation. InProceedings of the Advances in Neural Information Processing Systems, 2019

    work page 2019

  8. [7]

    Stop regressing: Training value functions via classification for scalable deep RL

    Jesse Farebrother, Jordi Orbay, Quan Vuong, Adrien Ali Taïga, Yevgen Chebotar, Ted Xiao, Alex Irpan, Sergey Levine, Pablo Samuel Castro, Aleksandra Faust, et al. Stop regressing: Training value functions via classification for scalable deep RL. InProceedings of the International Conference on Machine Learning, pages 13049–13071, 2024

    work page 2024

Show all 35 references
  1. [8]

    Shortcut learning in deep neural networks.Nature Machine Intelligence, 2(11):665–673, 2020

    Robert Geirhos, Jörn-Henrik Jacobsen, Claudio Michaelis, Richard Zemel, Wieland Brendel, Matthias Bethge, and Felix A Wichmann. Shortcut learning in deep neural networks.Nature Machine Intelligence, 2(11):665–673, 2020

    2020

  2. [9]

    Mastering diverse control tasks through world models.Nature, pages 1–7, 2025

    Danijar Hafner, Jurgis Pasukonis, Jimmy Ba, and Timothy Lillicrap. Mastering diverse control tasks through world models.Nature, pages 1–7, 2025

    2025

  3. [10]

    EgoDex: Learning dexterous manipulation from large-scale egocentric video.arXiv preprint arXiv:2505.11709, 2025

    Ryan Hoque, Peide Huang, David J Yoon, Mouli Sivapurapu, and Jian Zhang. EgoDex: Learning dexterous manipulation from large-scale egocentric video.arXiv preprint arXiv:2505.11709, 2025

    2025 arXiv

  4. [11]

    FiRE: En- hancing MLLMs with fine-grained context learning for complex image retrieval

    Bohan Hou, Haoqiang Lin, Xuemeng Song, Haokun Wen, Meng Liu, Yupeng Hu, and Xiangyu Zhao. FiRE: En- hancing MLLMs with fine-grained context learning for complex image retrieval. InProceedings of the International ACM SIGIR Conference on Research and Development in Information ...

    2025

  5. [12]

    Physical Intelligence, Ali Amin, Raichelle Aniceto, Ashwin Balakrishna, Kevin Black, Ken Conley, Grace Connors, James Darpinian, Karan Dhabalia, Jared DiCarlo, et al.π∗ 0.6: A VLA that learns from experience.arXiv preprint arXiv:2511.14759, 2025

    2025 arXiv

  6. [13]

    Galaxea open-world dataset and G0 dual-system VLA model.arXiv preprint arXiv:2509.00576, 2025

    Tao Jiang, Tianyuan Yuan, Yicheng Liu, Chenhao Lu, Jianning Cui, Xiao Liu, Shuiqi Cheng, Jiyang Gao, Huazhe Xu, and Hang Zhao. Galaxea open-world dataset and G0 dual-system VLA model.arXiv preprint arXiv:2509.00576, 2025

    2025 arXiv

  7. [14]

    Offline reinforcement learning with implicit Q-learning

    Ilya Kostrikov, Ashvin Nair, and Sergey Levine. Offline reinforcement learning with implicit Q-learning. In Proceedings of the International Conference on Learning Representations, 2022

    2022

  8. [15]

    RoboReward: General-purpose vision-language reward models for robotics.arXiv preprint arXiv:2601.00675, 2026

    Tony Lee, Andrew Wagenmaker, Karl Pertsch, Percy Liang, Sergey Levine, and Chelsea Finn. RoboReward: General-purpose vision-language reward models for robotics.arXiv preprint arXiv:2601.00675, 2026

    2026

  9. [16]

    Robometer: Scaling general-purpose robotic reward models via trajectory comparisons.arXiv preprint arXiv:2603.02115, 2026

    Anthony Liang, Yigit Korkmaz, Jiahui Zhang, Minyoung Hwang, Abrar Anwar, Sidhant Kaushik, Aditya Shah, Alex S Huang, Luke Zettlemoyer, Dieter Fox, et al. Robometer: Scaling general-purpose robotic reward models via trajectory comparisons.arXiv preprint arXiv:2603.02115, 2026

    2026 arXiv

  10. [17]

    RDT-1B: a diffusion foundation model for bimanual manipulation

    Songming Liu, Lingxuan Wu, Bangguo Li, Hengkai Tan, Huayu Chen, Zhengyi Wang, Ke Xu, Hang Su, and Jun Zhu. RDT-1B: a diffusion foundation model for bimanual manipulation. InProceedings of the International Conference on Learning Representations, pages 29982–30009, 2025

    2025

  11. [19]

    Vision language models are in-context value learners

    Yecheng Jason Ma, Joey Hejna, Ayzaan Wahid, Chuyuan Fu, Dhruv Shah, Jacky Liang, Zhuo Xu, Sean Kirmani, Peng Xu, Danny Driess, Ted Xiao, Jonathan Tompson, Osbert Bastani, Dinesh Jayaraman, Wenhao Yu, Tingnan Zhang, Dorsa Sadigh, and Fei Xia. Vision language models are in-conte...

    2025

  12. [20]

    RoboTwin: Dual-arm robot benchmark with generative digital twins

    Yao Mu, Tianxing Chen, Zanxin Chen, Shijia Peng, Zhiqian Lan, Zeyu Gao, Zhixuan Liang, Qiaojun Yu, Yude Zou, Mingkun Xu, et al. RoboTwin: Dual-arm robot benchmark with generative digital twins. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition...

    2025

  13. [21]

    Bigger, Regularized, Optimistic: scaling for compute and sample efficient continuous control

    Michal Nauman, Mateusz Ostaszewski, Krzysztof Jankowski, Piotr Mił oś, and Marek Cygan. Bigger, Regularized, Optimistic: scaling for compute and sample efficient continuous control. InProceedings of the Advances in Neural Information Processing Systems, pages 113038–113071, 2024

    2024

  14. [22]

    Policy invariance under reward transformations: Theory and application to reward shaping

    Andrew Y Ng, Daishi Harada, and Stuart Russell. Policy invariance under reward transformations: Theory and application to reward shaping. InProceedings of the International Conference on Machine Learning, pages 278–287, 1999

    1999

  15. [23]

    Open X-Embodiment: Robotic learning datasets and RT-X models

    Abby O’Neill, Abdul Rehman, Abhiram Maddukuri, Abhishek Gupta, and et.al. Open X-Embodiment: Robotic learning datasets and RT-X models. InProceedings of the IEEE International Conference on Robotics and Automation, pages 6892–6903, 2023

    2023

  16. [24]

    John Wiley & Sons, 2014

    Martin L Puterman.Markov decision processes: discrete stochastic dynamic programming. John Wiley & Sons, 2014

    2014

  17. [25]

    Universal value function approximators

    Tom Schaul, Daniel Horgan, Karol Gregor, and David Silver. Universal value function approximators. In Proceedings of the International Conference on Machine Learning, pages 1312–1320, 2015

    2015

  18. [26]

    Robo-Dopamine: General process reward modeling for high-precision robotic manipulation.arXiv preprint arXiv:2512.23703, 2025

    Huajie Tan, Sixiang Chen, Yijie Xu, Zixiao Wang, Yuheng Ji, Cheng Chi, Yaoxu Lyu, Zhongxia Zhao, Xian- sheng Chen, Peterson Co, et al. Robo-Dopamine: General process reward modeling for high-precision robotic manipulation.arXiv preprint arXiv:2512.23703, 2025

    2025

  19. [27]

    InternData-A1: Pioneering high-fidelity synthetic data for pre-training generalist policy

    Yang Tian, Yuyin Yang, Yiman Xie, Zetao Cai, Xu Shi, Ning Gao, Hangxu Liu, Xuekun Jiang, Zherui Qiu, Feng Yuan, et al. InternData-A1: Pioneering high-fidelity synthetic data for pre-training generalist policy. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pa...

    2026

  20. [28]

    Steering your diffusion policy with latent space reinforcement learning

    Andrew Wagenmaker, Mitsuhiko Nakamoto, Yunchu Zhang, Seohong Park, Waleed Yagoub, Anusha Nagabandi, Abhishek Gupta, and Sergey Levine. Steering your diffusion policy with latent space reinforcement learning. In Proceedings of the Conference on Robot Learning, 2025

    2025

  21. [29]

    RoboMIND: Benchmark on multi-embodiment intelligence normative data for robot manipulation.arXiv preprint arXiv:2412.13877, 2024

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

    2024 arXiv

  22. [30]

    RoboCOIN: An open-sourced bimanual robotic data collection for integrated manipulation

    Shihan Wu, Xuecheng Liu, Shaoxuan Xie, Pengwei Wang, Xinghang Li, Bowen Yang, Zhe Li, Kai Zhu, Hongyu Wu, Yiheng Liu, et al. RoboCOIN: An open-sourced bimanual robotic data collection for integrated manipulation. arXiv preprint arXiv:2511.17441, 2025

    2025 arXiv

  23. [31]

    RARM: Confidence-gated progress reward modeling for RL in manipulation.arXiv preprint arXiv:2606.22027, 2026

    Pengzhi Yang, Xinyu Wang, Pengyu Jing, Kehan Wen, Yiduo Qu, Zhenhao Huang, Minghao Fu, Xin Liu, Yaheng Shen, and Fan Shi. RARM: Confidence-gated progress reward modeling for RL in manipulation.arXiv preprint arXiv:2606.22027, 2026

    2026 arXiv

  24. [32]

    Zhilin Yang, Zihang Dai, Yiming Yang, Jaime Carbonell, Ruslan Salakhutdinov, and Quoc V. Le. XLNet: Generalized autoregressive pretraining for language understanding. InProceedings of the Advances in Neural Information Processing Systems, 2019

    2019

  25. [33]

    ReWiND: Language-guided rewards teach robot policies without new demonstrations

    Jiahui Zhang, Yusen Luo, Abrar Anwar, Sumedh Anand Sontakke, Joseph J Lim, Jesse Thomason, Erdem Biyik, and Jesse Zhang. ReWiND: Language-guided rewards teach robot policies without new demonstrations. In Proceedings of the Conference on Robot Learning, 2025

    2025

  26. [34]

    A generalist pair-wise progress critic model for vision-language-action robots

    Qi Zhang, Shaopeng Zhai, Shengzhe Zhang, Litao Liu, Tianyi Zhang, Fuxian Huang, and Ming Zhou. A generalist pair-wise progress critic model for vision-language-action robots. InProceedings of the International Conference on Machine Learning, 2026

    2026

  27. [35]

    Put the two pieces of bread in the basket

    Jinliang Zheng, Jianxiong Li, Zhihao Wang, Dongxiu Liu, Xirui Kang, Yuchun Feng, Yinan Zheng, Jiayin Zou, Yilun Chen, Jia Zeng, et al. X-VLA: Soft-prompted transformer as scalable cross-embodiment vision-language- action model.arXiv preprint arXiv:2510.10274, 2025. Appendix A ...

    arXiv

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.