pith. machine review for the scientific record. sign in

arxiv: 2604.27792 · v2 · submitted 2026-04-30 · 💻 cs.RO

Recognition: unknown

MotuBrain: An Advanced World Action Model for Robot Control

MotuBrain Team , Chendong Xiang , Fan Bao , Haitian Liu , Hengkai Tan , Hongzhe Bi , James Li , Jiabao Liu
show 12 more authors
Jingrui Pang Kiro Jing Louis Liu Mengchen Cai Rongxu Cui Ruowen Zhao Runqing Wang Shuhe Huang Yao Feng Yinze Rong Zeyuan Wang Jun Zhu
Authors on Pith no claims yet

Pith reviewed 2026-05-07 05:39 UTC · model grok-4.3

classification 💻 cs.RO
keywords roboticsworld modelsvision-language-actiondiffusion modelstransformer architecturerobot controlmultimodal learning
0
0 comments X

The pith

A single unified model achieves 95.8 percent success on complex robot tasks while adapting quickly to new embodiments.

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

MotuBrain is presented as a unified World Action Model that uses a UniDiffuser formulation with a three-stream Mixture-of-Transformers to model video and action together. This single model handles policy learning, world modeling, video generation, and other tasks while training on diverse data like video-only and cross-embodiment robot data. It reaches 95.8 percent success on clean settings and 96.1 percent on randomized settings on robot benchmarks, and adapts to new humanoid robots using just 50 to 100 trajectories. This matters because it suggests unified models can address the lack of fine-grained dynamics in existing vision-language-action approaches for practical robot deployment.

Core claim

The central discovery is that a three-stream Mixture-of-Transformers architecture under a UniDiffuser formulation enables a single model to jointly model video and action sequences, supporting multiple capabilities including policy learning and world modeling, while scaling to heterogeneous multimodal data and delivering 95.8 percent and 96.1 percent average success on robot benchmarks under clean and randomized settings respectively, along with strong comparative performance and efficient adaptation to new embodiments with 50 to 100 trajectories.

What carries the argument

three-stream Mixture-of-Transformers architecture under a UniDiffuser formulation that jointly processes multimodal streams for video, text, and action prediction

If this is right

  • Supports multiple functions such as policy learning, world modeling, video generation, inverse dynamics, and joint video-action prediction from a single model.
  • Scales effectively to heterogeneous data including video-only, task-agnostic, and cross-embodiment robot data.
  • Achieves over 50x speedup in inference through optimizations like quantization and caching, enabling real-time control up to 11 Hz.
  • Delivers robust performance on both clean and randomized benchmark settings.

Where Pith is reading between the lines

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

  • This approach may allow for more efficient development of robot systems by eliminating the need for separate models for different capabilities.
  • Future work could test integration with additional sensor modalities to further enhance long-horizon control in unpredictable environments.
  • The shared cross-embodiment action representation might facilitate transfer learning across a wider range of robot platforms beyond those tested.

Load-bearing premise

The three-stream Mixture-of-Transformers architecture under the UniDiffuser formulation, along with the training data mixtures and post-training optimizations, directly produces the reported performance improvements without relying on benchmark-specific tuning or overfitting that fails to generalize.

What would settle it

Running the model on a novel robot task or embodiment outside the training distribution and measuring if success rates remain above 90 percent with only 50-100 trajectories would provide a direct test of the adaptability claim.

read the original abstract

Vision-Language-Action (VLA) models generalize semantically well but often lack fine-grained modeling of world dynamics. We present MotuBrain, a unified World Action Model that jointly models video and action under a UniDiffuser formulation with a three-stream Mixture-of-Transformers architecture. A single model supports policy learning, world modeling, video generation, inverse dynamics, and joint video-action prediction, while scaling to heterogeneous multimodal data such as video-only, task-agnostic, and cross-embodiment robot data. Building on Motus, MotuBrain further introduces unified multiview modeling, an independent text stream for stronger language-action coupling, a shared cross-embodiment action representation, and an efficient post-training and deployment recipe for long-horizon real-world control. Our inference stack combines step reduction, compilation, FP8 quantization, DiT caching, V2A-style action-only inference, and real-time chunked closed-loop execution, achieving over 50x speedup over a naive baseline and up to 11 Hz inference. Experimentally, MotuBrain achieves 95.8% and 96.1% average success on RoboTwin 2.0 under clean and randomized settings, respectively, attains the strongest reported EWMScore in our WorldArena comparison, and adapts to new humanoid embodiments with only 50--100 trajectories. These results show that unified world action models can scale in generality, predictive accuracy, and real-world deployability.

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

3 major / 2 minor

Summary. MotuBrain is presented as a unified World Action Model that jointly models video and action under a UniDiffuser formulation using a three-stream Mixture-of-Transformers architecture. A single model supports policy learning, world modeling, video generation, inverse dynamics, and joint video-action prediction while scaling to heterogeneous data including video-only, task-agnostic, and cross-embodiment robot trajectories. The work introduces unified multiview modeling, an independent text stream, a shared cross-embodiment action representation, and an efficient post-training and deployment recipe. Experimentally, it reports 95.8% and 96.1% average success on RoboTwin 2.0 under clean and randomized settings, the strongest EWMScore in a WorldArena comparison, and adaptation to new humanoid embodiments using only 50-100 trajectories, together with an inference stack (step reduction, FP8, DiT caching, V2A-style inference) achieving over 50x speedup and up to 11 Hz real-time control.

Significance. If the performance claims are substantiated with proper controls, this would constitute a meaningful contribution to robot learning by demonstrating that unified world-action models can simultaneously achieve high task generality, predictive accuracy, and practical real-world deployability. The multi-task formulation and the emphasis on efficient inference for long-horizon control address important gaps between current VLA models and deployable systems. The reported few-shot adaptation to new embodiments is particularly noteworthy if shown to generalize beyond the specific benchmarks.

major comments (3)
  1. [Experimental Results] The manuscript reports 95.8% and 96.1% average success on RoboTwin 2.0 (clean and randomized) and the strongest EWMScore on WorldArena, yet supplies no experimental details on baselines, number of evaluation runs, error bars, data splits, or statistical significance. This omission is load-bearing for the central empirical claims because the abstract and results cannot be verified or compared to prior VLA work without these elements.
  2. [Architecture and Ablations] The paper attributes the performance gains and few-shot adaptation to the three-stream Mixture-of-Transformers architecture under UniDiffuser together with the independent text stream and shared cross-embodiment action representation. However, no ablation studies are described that remove the text stream or the cross-embodiment representation while holding the training data mixtures and post-training optimizations (step reduction, FP8, DiT caching) fixed. Without such controls, it is impossible to isolate whether the architecture, rather than data curation or the inference recipe, drives the reported 95.8%/96.1% success and 50-100-trajectory adaptation.
  3. [Model Capabilities] The claim that a single model supports policy learning, world modeling, video generation, inverse dynamics, and joint prediction is central to the unified-model narrative, yet the text provides no quantitative results or task-specific metrics demonstrating simultaneous competence across these capabilities on the same model checkpoint.
minor comments (2)
  1. [Abstract and Introduction] The abstract and introduction use several acronyms (VLA, EWMScore, UniDiffuser, DiT, V2A) without first defining them for readers outside the immediate subfield.
  2. [Figures and Tables] Figure captions and table headers should explicitly state the number of runs and whether error bars represent standard deviation or standard error.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We sincerely thank the referee for the constructive feedback and for recognizing the potential significance of our work. We address each major comment below and outline the revisions we will make to strengthen the manuscript.

read point-by-point responses

  1. Referee: [Experimental Results] The manuscript reports 95.8% and 96.1% average success on RoboTwin 2.0 (clean and randomized) and the strongest EWMScore on WorldArena, yet supplies no experimental details on baselines, number of evaluation runs, error bars, data splits, or statistical significance. This omission is load-bearing for the central empirical claims because the abstract and results cannot be verified or compared to prior VLA work without these elements.

    Authors: We fully agree that these details are crucial for validating our claims. In the revised manuscript, we will include a detailed experimental setup section specifying the baselines (including their original papers and implementations), the number of evaluation runs (10 runs with different random seeds for each task), error bars (standard deviation across runs), data splits (e.g., training on 80% of trajectories and testing on 20%), and statistical significance (using t-tests to compare against baselines). This will allow proper verification and comparison. revision: yes

  2. Referee: [Architecture and Ablations] The paper attributes the performance gains and few-shot adaptation to the three-stream Mixture-of-Transformers architecture under UniDiffuser together with the independent text stream and shared cross-embodiment action representation. However, no ablation studies are described that remove the text stream or the cross-embodiment representation while holding the training data mixtures and post-training optimizations (step reduction, FP8, DiT caching) fixed. Without such controls, it is impossible to isolate whether the architecture, rather than data curation or the inference recipe, drives the reported 95.8%/96.1% success and 50-100-trajectory adaptation.

    Authors: We recognize the need to isolate the contributions of the architectural components. To address this, we will add ablation experiments in the revision. We will train and evaluate variants of the model without the independent text stream and without the shared cross-embodiment action representation, while keeping the training data mixtures and post-training optimizations identical. Performance on RoboTwin 2.0 and adaptation tasks will be reported to quantify the impact. Due to computational constraints, these ablations will be performed on a representative subset of tasks, with a discussion of the results. revision: partial

  3. Referee: [Model Capabilities] The claim that a single model supports policy learning, world modeling, video generation, inverse dynamics, and joint prediction is central to the unified-model narrative, yet the text provides no quantitative results or task-specific metrics demonstrating simultaneous competence across these capabilities on the same model checkpoint.

    Authors: We agree that demonstrating multi-capability with quantitative metrics on the same checkpoint is important to support the unified model claim. In the revised manuscript, we will add a new table and section providing quantitative results for each capability using the same MotuBrain checkpoint. This will include: policy learning success rates on RoboTwin, world modeling metrics such as video prediction accuracy (e.g., MSE or PSNR), video generation quality (FID, CLIP score), inverse dynamics prediction accuracy, and joint video-action prediction metrics. These evaluations will be on held-out data to show the model's versatility without task-specific fine-tuning. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical results with no derivation chain

full rationale

The paper describes an architecture (three-stream Mixture-of-Transformers under UniDiffuser) and reports empirical success rates on RoboTwin 2.0 and WorldArena without any mathematical derivation, equations, or first-principles predictions. All performance claims are measured outcomes from training and inference on specific data mixtures, not quantities derived by construction from fitted parameters or self-referential definitions. The reference to building on 'Motus' is a high-level architectural extension rather than a load-bearing self-citation that reduces the central claims to unverified priors. No steps match the enumerated circularity patterns, so the result is self-contained empirical reporting.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

The abstract provides no explicit mathematical derivations, free parameters, or axioms; the central claims rest on the empirical effectiveness of the described architecture and data mixture, whose details are not supplied here. No invented entities with independent evidence are introduced in the text.

pith-pipeline@v0.9.0 · 5628 in / 1283 out tokens · 45644 ms · 2026-05-07T05:39:13.885899+00:00 · methodology

discussion (0)

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

Forward citations

Cited by 1 Pith paper

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

  1. World Action Models: The Next Frontier in Embodied AI

    cs.RO 2026-05 unverdicted novelty 4.0

    The paper introduces World Action Models as a new paradigm unifying predictive world modeling with action generation in embodied foundation models and provides a taxonomy of existing approaches.

Reference graph

Works this paper leans on

41 extracted references · 13 canonical work pages · cited by 1 Pith paper · 9 internal anchors

  1. [1]

    Qwen3-VL Technical Report

    Shuai Bai, Yuxuan Cai, Ruizhe Chen, Keqin Chen, Xionghui Chen, Zesen Cheng, Lianghao Deng, Wei Ding, Chang Gao, Chunjiang Ge, et al. Qwen3-vl technical report. arXiv preprint arXiv:2511.21631, 2025

    work page internal anchor Pith review arXiv 2025

  2. [2]

    One transformer fits all distributions in multi-modal diffusion at scale, 2023

    Fan Bao, Shen Nie, Kaiwen Xue, Chongxuan Li, Shi Pu, Yaole Wang, Gang Yue, Yue Cao, Hang Su, and Jun Zhu. One transformer fits all distributions in multi-modal diffusion at scale, 2023

    2023

  3. [3]

    Vidu: a highly consistent, dynamic and skilled text-to-video generator with diffusion models, 2024

    Fan Bao, Chendong Xiang, Gang Yue, Guande He, Hongzhou Zhu, Kaiwen Zheng, Min Zhao, Shilong Liu, Yaole Wang, and Jun Zhu. Vidu: a highly consistent, dynamic and skilled text-to-video generator with diffusion models, 2024

    2024

  4. [4]

    PaliGemma: A versatile 3B VLM for transfer

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

    work page internal anchor Pith review arXiv 2024

  5. [5]

    Motus: A unified latent action world model, 2025

    Hongzhe Bi, Hengkai Tan, Shenghao Xie, Zeyuan Wang, Shuhe Huang, Haitian Liu, Ruowen Zhao, Yao Feng, Chendong Xiang, Yinze Rong, Hongyan Zhao, Hanyu Liu, Zhizhong Su, Lei Ma, Hang Su, and Jun Zhu. Motus: A unified latent action world model, 2025

    2025

  6. [6]

    Kevin Black, Noah Brown, Danny Driess, Adnan Esmail, Michael Equi, Chelsea Finn, Niccolo Fusai, Lachy Groom, Karol Hausman, Brian Ichter, Szymon Jakubczak, Tim Jones, Liyiming Ke, Sergey Levine, Adrian Li-Bell, Mohith Mothukuri, Suraj Nair, Karl Pertsch, Lucy Xiaoyang Shi, James Tanner, Quan Vuong, Anna Walling, Haohuan Wang, and Ury Zhilinsky .π0: A visi...

    2026

  7. [7]

    Y ., and Levine, S

    Kevin Black, Manuel Y Galliker, and Sergey Levine. Real-time execution of action chunking flow policies. arXiv preprint arXiv:2506.07339, 2025

    work page arXiv 2025

  8. [8]

    Rt-2: Vision-language-action models transfer web knowledge to robotic control, 2023.18

    Anthony Brohan, Noah Brown, Justice Carbajal, Yevgen Chebotar, Xi Chen, Krzysztof Choromanski, Tianli Ding, Danny Driess, Avinava Dubey, Chelsea Finn, Pete Florence, Chuyuan Fu, Montse Gonzalez Arenas, Keerthana Gopalakrishnan, Kehang Han, Karol Hausman, Alexander Herzog, Jasmine Hsu, Brian Ichter, Alex Irpan, Nikhil Joshi, Ryan Julian, Dmitry Kalashnikov...

    2023

  9. [9]

    Genie: Generative interactive environments, 2024

    Jake Bruce, Michael Dennis, Ashley Edwards, Jack Parker-Holder, Yuge Shi, Edward Hughes, Matthew Lai, Aditi Mavalankar, Richie Steigerwald, Chris Apps, Yusuf Aytar, Sarah Bechtle, Feryal Behbahani, Stephanie Chan, Nicolas Heess, Lucy Gonzalez, Simon Osindero, Sherjil Ozair, Scott Reed, Jingwei Zhang, Konrad Zolna, Jeff Clune, Nando de Freitas, Satinder Si...

    2024

  10. [10]

    Univla: Learning to act anywhere with task-centric latent actions, 2025

    Qingwen Bu, Yanting Yang, Jisong Cai, Shenyuan Gao, Guanghui Ren, Maoqing Yao, Ping Luo, and Hongyang Li. Univla: Learning to act anywhere with task-centric latent actions, 2025

    2025

  11. [11]

    RoboTwin 2.0: A Scalable Data Generator and Benchmark with Strong Domain Randomization for Robust Bimanual Robotic Manipulation

    Tianxing Chen, Zanxin Chen, Baijun Chen, Zijian Cai, Yibin Liu, Zixuan Li, Qiwei Liang, Xianliang Lin, Yiheng Ge, Zhenyu 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 arXiv 2025

  12. [12]

    Open x-embodiment: Robotic learning datasets and rt-x models, 2023

    Open X-Embodiment Collaboration. Open x-embodiment: Robotic learning datasets and rt-x models, 2023

    2023

  13. [13]

    Tenenbaum, Dale Schuurmans, and Pieter Abbeel

    Yilun Du, Mengjiao Yang, Bo Dai, Hanjun Dai, Ofir Nachum, Joshua B. Tenenbaum, Dale Schuurmans, and Pieter Abbeel. Learning universal policies via text-guided video generation, 2023

    2023

  14. [14]

    Vidar: Embodied video diffusion model for generalist manipulation.arXiv preprint arXiv:2507.12898, 2025

    Yao Feng, Hengkai Tan, Xinyi Mao, Chendong Xiang, Guodong Liu, Shuhe Huang, Hang Su, and Jun Zhu. Vidar: Embodied video diffusion model for generalist manipulation. arXiv preprint arXiv:2507.12898, 2025

    work page arXiv 2025

  15. [15]

    Adaworld: Learning adaptable world models with latent actions, 2025

    Shenyuan Gao, Siyuan Zhou, Yilun Du, Jun Zhang, and Chuang Gan. Adaworld: Learning adaptable world models with latent actions, 2025

    2025

  16. [16]

    Ctrl-world: A controllable generative world model for robot manipulation, 2025

    Yanjiang Guo, Lucy Xiaoyang Shi, Jianyu Chen, and Chelsea Finn. Ctrl-world: A controllable generative world model for robot manipulation. arXiv preprint arXiv:2510.10125, 2025

    work page arXiv 2025

  17. [17]

    Pre-trained video generative models as world simulators, 2025

    Haoran He, Yang Zhang, Liang Lin, Zhongwen Xu, and Ling Pan. Pre-trained video generative models as world simulators, 2025

    2025

  18. [18]

    Matrix-game 2.0: An open-source real-time and streaming interactive world model, 2026

    Xianglong He, Chunli Peng, Zexiang Liu, Boyang Wang, Yifan Zhang, Qi Cui, Fei Kang, Biao Jiang, Mengyin An, Yangyang Ren, Baixin Xu, Hao-Xiang Guo, Kaixiong Gong, Size Wu, Wei Li, Xuchen Song, Yang Liu, Yangguang Li, and Yahui Zhou. Matrix-game 2.0: An open-source real-time and streaming interactive world model, 2026

    2026

  19. [19]

    Video Prediction Policy: A Generalist Robot Policy with Predictive Visual Representations

    Yucheng Hu, Yanjiang Guo, Pengchao Wang, Xiaoyu Chen, Yen-Jen Wang, Jianke Zhang, Koushil Sreenath, Chaochao Lu, and Jianyu Chen. Video prediction policy: A generalist robot policy with predictive visual representations. arXiv preprint arXiv:2412.14803, 2024

    work page internal anchor Pith review arXiv 2024

  20. [20]

    Physical Intelligence, Kevin Black, Noah Brown, James Darpinian, Karan Dhabalia, Danny Driess, Adnan Esmail, Michael Equi, Chelsea Finn, Niccolo Fusai, Manuel Y. Galliker, Dibya Ghosh, Lachy Groom, Karol Hausman, Brian Ichter, Szymon Jakubczak, Tim Jones, Liyiming Ke, Devin LeBlanc, Sergey Levine, Adrian Li-Bell, Mohith Mothukuri, Suraj Nair, Karl Pertsch...

    2025

  21. [21]

    Cosmos Policy: Fine-Tuning Video Models for Visuomotor Control and Planning

    Moo Jin Kim, Yihuai Gao, Tsung-Yi Lin, Yen-Chen Lin, Yunhao Ge, Grace Lam, Percy Liang, Shuran Song, Ming-Yu Liu, Chelsea Finn, et al. Cosmos policy: Fine-tuning video models for visuomotor control and planning. arXiv preprint arXiv:2601.16163, 2026

    work page internal anchor Pith review arXiv 2026

  22. [22]

    Openvla: An open-source vision-language-action model, 2024

    Moo Jin Kim, Karl Pertsch, Siddharth Karamcheti, Ted Xiao, Ashwin Balakrishna, Suraj Nair, Rafael Rafailov, Ethan Foster, Grace Lam, Pannag Sanketi, Quan Vuong, Thomas Kollar, Benjamin Burchfiel, Russ Tedrake, Dorsa Sadigh, Sergey Levine, Percy Liang, and Chelsea Finn. Openvla: An open-source vision-language-action model, 2024

    2024

  23. [23]

    Hunyuanvideo: A systematic framework for large video generative models, 2025

    Weijie Kong, Qi Tian, Zijian Zhang, Rox Min, Zuozhuo Dai, Jin Zhou, Jiangfeng Xiong, Xin Li, Bo Wu, Jianwei Zhang, Kathrina Wu, Qin Lin, Junkun Yuan, Yanxin Long, Aladdin Wang, Andong Wang, Changlin Li, Duojun Huang, Fang Yang, Hao Tan, Hongmei Wang, Jacob Song, Jiawang Bai, Jianbing Wu, Jinbao Xue, Joey Wang, Kai Wang, Mengyang Liu, Pengyu Li, Shuai Li, ...

    2025

  24. [24]

    Causal World Modeling for Robot Control

    Lin Li, Qihang Zhang, Yiming Luo, Shuai Yang, Ruilin Wang, Fei Han, Mingrui Yu, Zelin Gao, Nan Xue, Xing Zhu, et al. Causal world modeling for robot control. arXiv preprint arXiv:2601.21998, 2026

    work page internal anchor Pith review arXiv 2026

  25. [25]

    Genieenvisioner: A unified world foundation platform for robotic manipulation.arXiv preprint arXiv:2508.05635,

    Yue Liao, Pengfei Zhou, Siyuan Huang, Donglin Yang, Shengcong Chen, Yuxin Jiang, Yue Hu, Jingbin Cai, Si Liu, Jianlan Luo, Liliang Chen, Shuicheng Yan, Maoqing Yao, and Guanghui Ren. Genie envisioner: A unified world foundation platform for robotic manipulation. arXiv preprint arXiv:2508.05635, 2025. 19

    work page arXiv 2025

  26. [26]

    Rdt-1b: a diffusion foundation model for bimanual manipulation, 2025

    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, 2025

    2025

  27. [27]

    Lda-1b: Scaling latent dynamics action model via universal embodied data ingestion, 2026

    Jiangran Lyu, Kai Liu, Xuheng Zhang, Haoran Liao, Yusen Feng, Wenxuan Zhu, Tingrui Shen, Jiayi Chen, Jiazhao Zhang, Yifei Dong, Wenbo Cui, Senmao Qi, Shuo Wang, Yixin Zheng, Mi Yan, Xuesong Shi, Haoran Li, Dongbin Zhao, Ming-Yu Liu, Zhizheng Zhang, Li Yi, Yizhou Wang, and He Wang. Lda-1b: Scaling latent dynamics action model via universal embodied data in...

    2026

  28. [28]

    Cosmos world foundation model platform for physical ai, 2025

    NVIDIA, :, Niket Agarwal, Arslan Ali, Maciej Bala, Yogesh Balaji, Erik Barker, Tiffany Cai, Prithvijit Chattopadhyay, Yongxin Chen, Yin Cui, Yifan Ding, Daniel Dworakowski, Jiaojiao Fan, Michele Fenzi, Francesco Ferroni, Sanja Fidler, Dieter Fox, Songwei Ge, Yunhao Ge, Jinwei Gu, Siddharth Gururani, Ethan He, Jiahui Huang, Jacob Huffman, Pooya Jannaty, Ji...

    2025

  29. [29]

    mimic-video: Video-action models for generalizable robot control beyond vlas, 2025

    Jonas Pai, Liam Achenbach, Victoriano Montesinos, Benedek Forrai, Oier Mees, and Elvis Nava. mimic-video: Video-action models for generalizable robot control beyond vlas, 2025

    2025

  30. [30]

    Team Seedance, De Chen, Liyang Chen, Xin Chen, Ying Chen, Zhuo Chen, Zhuowei Chen, Feng Cheng, Tianheng Cheng, Yufeng Cheng, Mojie Chi, Xuyan Chi, Jian Cong, Qinpeng Cui, Fei Ding, Qide Dong, Yujiao Du, Haojie Duanmu, Junliang Fan, Jiarui Fang, Jing Fang, Zetao Fang, Chengjian Feng, Yu Gao, Diandian Gu, Dong Guo, Hanzhong Guo, Qiushan Guo, Boyang Hao, Hon...

    2026

  31. [31]

    Worldarena: A unified benchmark for evaluating perception and functional utility of embodied world models, 2026

    Yu Shang, Zhuohang Li, Yiding Ma, Weikang Su, Xin Jin, Ziyou Wang, Lei Jin, Xin Zhang, Yinzhou Tang, Haisheng Su, et al. Worldarena: A unified benchmark for evaluating perception and functional utility of embodied world models, 2026

    2026

  32. [32]

    Anypos: Automated task-agnostic actions for bimanual manipulation, 2025

    Hengkai Tan, Yao Feng, Xinyi Mao, Shuhe Huang, Guodong Liu, Zhongkai Hao, Hang Su, and Jun Zhu. Anypos: Automated task-agnostic actions for bimanual manipulation, 2025

    2025

  33. [33]

    Octo: An open-source generalist robot policy , 2024

    Octo Model Team, Dibya Ghosh, Homer Walke, Karl Pertsch, Kevin Black, Oier Mees, Sudeep Dasari, Joey Hejna, Tobias Kreiman, Charles Xu, Jianlan Luo, You Liang Tan, Lawrence Yunliang Chen, Pannag Sanketi, Quan Vuong, Ted Xiao, Dorsa Sadigh, Chelsea Finn, and Sergey Levine. Octo: An open-source generalist robot policy , 2024

    2024

  34. [34]

    Wan: Open and Advanced Large-Scale Video Generative Models

    Team Wan, Ang Wang, Baole Ai, Bin Wen, Chaojie Mao, Chen-Wei Xie, Di Chen, Feiwu Yu, Haiming Zhao, Jianxiao Yang, Jianyuan Zeng, Jiayu Wang, Jingfeng Zhang, Jingren Zhou, Jinkai Wang, Jixuan Chen, Kai Zhu, Kang Zhao, Keyu Yan, Lianghua Huang, Mengyang Feng, Ningyi Zhang, Pandeng Li, Pingyu Wu, Ruihang Chu, Ruili Feng, Shiwei Zhang, Siyang Sun, Tao Fang, T...

    work page internal anchor Pith review arXiv 2025

  35. [35]

    Hbridge: H-shape bridging of heterogeneous experts for unified multimodal understanding and generation, 2025

    Xiang Wang, Zhifei Zhang, He Zhang, Zhe Lin, Yuqian Zhou, Qing Liu, Shiwei Zhang, Yijun Li, Shaoteng Liu, Haitian Zheng, Jason Kuen, Yuehuan Wang, Changxin Gao, and Nong Sang. Hbridge: H-shape bridging of heterogeneous experts for unified multimodal understanding and generation, 2025

    2025

  36. [36]

    World Action Models are Zero-shot Policies

    Seonghyeon Ye, Yunhao Ge, Kaiyuan Zheng, Shenyuan Gao, Sihyun Yu, George Kurian, Suneel Indupuru, You Liang Tan, Chuning Zhu, Jiannan Xiang, et al. World action models are zero-shot policies.arXiv preprint arXiv:2602.15922, 2026

    work page internal anchor Pith review arXiv 2026

  37. [37]

    Fast-WAM: Do World Action Models Need Test-time Future Imagination?

    Tianyuan Yuan, Zibin Dong, Yicheng Liu, and Hang Zhao. Fast-wam: Do world action models need test-time future imagination? arXiv preprint arXiv:2603.16666, 2026

    work page internal anchor Pith review arXiv 2026

  38. [38]

    X-vla: Soft-prompted transformer as scalable cross-embodiment vision-language-action model, 2025

    Jinliang Zheng, Jianxiong Li, Zhihao Wang, Dongxiu Liu, Xirui Kang, Yuchun Feng, Yinan Zheng, Jiayin Zou, Yilun Chen, Jia Zeng, Ya-Qin Zhang, Jiangmiao Pang, Jingjing Liu, Tai Wang, and Xianyuan Zhan. X-vla: Soft-prompted transformer as scalable cross-embodiment vision-language-action model, 2025

    2025

  39. [39]

    Open-sora: Democratizing efficient video production for all, 2024

    Zangwei Zheng, Xiangyu Peng, Tianji Yang, Chenhui Shen, Shenggui Li, Hongxin Liu, Yukun Zhou, Tianyi Li, and Yang You. Open-sora: Democratizing efficient video production for all, 2024

    2024

  40. [40]

    Robodreamer: Learning compositional world models for robot imagination, 2024

    Siyuan Zhou, Yilun Du, Jiaben Chen, Yandong Li, Dit-Yan Yeung, and Chuang Gan. Robodreamer: Learning compositional world models for robot imagination, 2024

    2024

  41. [41]

    Unified world models: Coupling video and action diffusion for pretraining on large robotic datasets

    Chuning Zhu, Raymond Yu, Siyuan Feng, Benjamin Burchfiel, Paarth Shah, and Abhishek Gupta. Unified world models: Coupling video and action diffusion for pretraining on large robotic datasets. In Proceedings of Robotics: Science and Systems (RSS), 2025. 21

    2025