Pith. sign in

REVIEW 3 major objections 5 minor 29 references

A dual-system robot policy keeps world-action execution as the default path and, under a matched real folding protocol, beats strong vision-language-action policies on success and speed.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · grok-4.5

2026-07-11 11:20 UTC pith:5Z22WZ2P

load-bearing objection Solid systems paper with a useful matched WAM-vs-VLA folding comparison; the dual-system packaging is mostly packaging, and the real-robot win is real but not airtight on optimization parity. the 3 major comments →

arxiv 2607.04927 v1 pith:5Z22WZ2P submitted 2026-07-06 cs.RO cs.AI

DSWAM: A Dual-System World Action Foundation Model for Fine-Grained Robot Manipulation

Jian Zhu , Jianjun Zhang , Taiyi Su , Tianbin Liu , Zhangyuan Wang , Kai Xie , Zitai Huang , Chong Ma
show 5 more authors
Youzhang He Tianjian Wang Hanyang Wang Weihao Ding Yi Xu
This is my paper
classification cs.RO cs.AI
keywords World Action ModelDual-SystemFine-Grained Robot ManipulationVideo Co-trainingTensorRT AccelerationReal-time ChunkingDeformable ManipulationVision-Language-Action
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

This paper argues that robot manipulation needs two separable skills: physically grounded execution and, only when useful, language-level decomposition of coarse household commands. It presents DSWAM, which keeps a world action model as the always-on low-level executor and adds an optional vision-language subtask planner that turns multi-step goals into fine-grained executable instructions. The executor is trained with action prediction plus video co-training so it learns how scenes evolve under contact, but at test time it predicts action chunks directly without generating future video. Under a carefully matched real-robot folding setup with the same platform, data, tasks, and scoring as a strong vision-language-action baseline, the world-action executor alone raises average success from 92.5% to 96.3% and cuts average completion time from 2 minutes 18 seconds to 1 minute 44 seconds. The optional planner further stabilizes tasks that truly need decomposition, and TensorRT plus asynchronous real-time chunking make the policy practical on hardware.

Core claim

Under a matched real-world deformable folding protocol with identical robot platform, pretraining data, post-training data, task definitions, and success criteria, a world-action executor trained with video co-training but run without future-video imagination outperforms a strong vision-language-action baseline, improving average success from 92.5% to 96.3% and reducing average completion time from 2′18″ to 1′44″, while an optional high-level subtask planner improves stability only when coarse commands benefit from decomposition.

What carries the argument

DSWAM’s dual-system split: System 1 is a world action model that encodes multi-view vision, language, and proprioception into latent world features and predicts continuous action chunks via flow matching, trained with joint action and video co-training but with no future-video generation at inference; System 2 is an optional vision-language planner that, only when activated, maps short visual history plus a global prompt into ordered executable subtasks that condition System 1.

Load-bearing premise

Matching robot, data, task protocol, and success criteria is enough to isolate world-action versus vision-language-action policy design, even without fully disclosed equal compute, hyperparameter search, and independent reimplementation under the same optimization budget.

What would settle it

Re-train both the world-action executor and the matched vision-language-action baseline from the same data with locked compute, seeds, and hyperparameter budgets on the same folding suite; if average success and completion time no longer favor the world-action path, the central execution claim fails.

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

If this is right

  • World-action models can serve as the default low-level controller for contact-rich household work without requiring high-level planning on every step.
  • Video co-training can supply physical supervision at train time while inference stays a direct action-chunk policy, avoiding costly future-frame generation on the robot.
  • Optional subtask language improves multi-step sorting-style tasks by cutting mistakes, without forcing planners into atomic commands.
  • TensorRT plus asynchronous real-time chunking can keep world-action policies inside real control timing on dual-arm platforms.
  • Matched real-robot protocols become the right standard for claiming world-action versus vision-language-action gains rather than cross-setup score tables.

Where Pith is reading between the lines

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

  • If train-time world modeling plus direct action inference is the real gain, many expensive imagine-then-act stacks could be simplified to action-only serving without losing contact-rich performance.
  • The dual-system split suggests household robots may need a lightweight ‘when to plan’ gate rather than always-on hierarchical planners.
  • Success on pants folding and randomized simulation may indicate the method helps most where object state is hard to read from a single frame.
  • A natural next test is long-horizon kitchen tasks with open-ended object sets, where planner error could compound even if low-level execution is strong.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper introduces DSWAM, a dual-system robot foundation model that keeps a System 1 World Action Model (WAM) executor as the default control path and optionally activates a System 2 vision-language subtask planner for coarse multi-step instructions. System 1 is trained with action prediction plus video co-training under a shared flow-matching objective (Eqs. 11–15) but at inference predicts action chunks without future-video generation; System 2 uses transition-aware subtask labels (Eq. 6) and is invoked only when decomposition is useful. The authors further integrate TensorRT, asynchronous execution, and real-time chunking for deployment. Empirically, under a matched DeMaVLA real-world folding protocol (same robot, pretraining/post-training data, tasks, and criteria) with System 2 disabled, DSWAM raises average success from 92.5% to 96.3% and cuts completion time from 2′18″ to 1′44″ (Table 2); it also reports strong RoboTwin 2.0 averages (Table 1), a sorting ablation for subtask supervision (Table 3), and latency/rollout gains for TensorRT+RTC (Tables 4–5).

Significance. If the matched-protocol gains hold under tighter optimization and statistical controls, the paper would provide one of the cleaner real-robot comparisons between WAM-style video-co-trained execution and strong VLA policies on contact-rich deformable manipulation, while also offering a practical dual-system design that keeps planning optional rather than mandatory. Strengths include the explicit WAM-only folding comparison, the transition-aware planner supervision, the no-future-generation inference path, and concrete deployment measurements (TensorRT BF16 2.69× speedup; asynchronous RTC rollouts). These are useful contributions for the robot foundation-model community even if the absolute SR margins remain modest.

major comments (3)
  1. §4.1 and §4.3 / Table 2: The central claim that WAM-style execution outperforms a matched VLA under identical robot, data, protocol, and criteria is load-bearing, but the manuscript does not report training compute, hyperparameter search, seed variance, confidence intervals, or an independent reimplementation of DeMaVLA under the same optimization budget. With only 20 trials per garment category and no uncertainty estimates, the 3.8-point average SR gain (and the pants-driven 75%→90% jump) could partly reflect training recipe, checkpoint selection, or the TensorRT/RTC stack rather than video co-training versus VLA modeling. Please add optimization-parity details and statistical uncertainty, or qualify the causal attribution.
  2. §4.5 / Tables 4–5 vs. Table 2: Asynchronous TensorRT+RTC is shown to improve pants success from 70% to 100% and to shorten success times on easy garments, yet the main folding comparison (Table 2) does not state whether π0 and DeMaVLA were evaluated under the same asynchronous RTC deployment stack. If they were not, part of the reported time and SR advantage may be system-level rather than policy-class. Clarify the deployment conditions for all methods in the matched benchmark, or ablate RTC/async off for DSWAM in Table 2.
  3. §3.2 and §4.4 / Table 3: System 2 is presented as an optional dual-system contribution, but the only real-robot evidence is a two-object sorting task with coarse vs. subtask-level instruction supervision of the executor, not a closed-loop evaluation of the full planner–executor protocol (Eq. 8, Δt=2 s) on multi-step household tasks. The folding benchmark deliberately disables System 2. Without a planner-on vs. planner-off study on a genuinely multi-step task, the dual-system claim remains only partially supported. A small closed-loop System 2 experiment (or clearer scope limitation) is needed.
minor comments (5)
  1. §3.3.1: Free parameters λ_vid, action horizon H, and the flow-matching noise schedule are not numerically specified; a short hyperparameter table would aid reproducibility.
  2. Table 1: Several tasks show large absolute swings across methods (e.g., Open Microwave, Hanging Mug); a brief discussion of failure modes would help interpret the 92.38%/91.90% averages.
  3. Figure 1 caption and §3.1: The dual-system diagram is clear, but the text should state more explicitly when System 2 is auto-triggered versus manually enabled at deployment.
  4. References: DeMaVLA is cited as arXiv:2605.31286 with substantial author overlap; a short note on independence of the baseline evaluation would reduce ambiguity for readers.
  5. Notation: Ot in Eq. (1) and the latent zt in Eq. (9) are clear, but the relationship between video latent tokens Vt+1:t+T and the Wan2.2 backbone could be stated more precisely.

Circularity Check

0 steps flagged

Empirical systems paper with no derivation that reduces to its inputs by construction; matched DeMaVLA comparison is experimental, not circular.

full rationale

DSWAM is a robotics systems/empirical paper, not a first-principles derivation. Its load-bearing claims are measured success rates and completion times on RoboTwin 2.0 and a real-world folding protocol (Tables 1–2), plus ablations on subtask supervision and TensorRT/RTC deployment (Tables 3–5). None of these quantities are defined in terms of the claimed outputs, fitted and then re-reported as predictions, or forced by a uniqueness theorem. Video co-training (Eqs. 11–15) is a standard multi-task training objective; inference deliberately drops future-video generation, so the training signal is not algebraically identical to the reported action policy. Author overlap with DeMaVLA [22] makes the matched baseline a same-lab comparison and raises ordinary methodological burden (optimization parity, N, variance), but that is not circularity: the paper does not justify “WAM beats VLA” by citing DeMaVLA as a theorem—it reports new rollouts under a shared protocol. Dual-system framing and RTC/TensorRT are architectural/deployment choices, not renamings of a fitted empirical law presented as derivation. No self-definitional loop, fitted-input-as-prediction, uniqueness import, or ansatz-via-self-citation chain is present. Score 0 is the honest finding.

Axiom & Free-Parameter Ledger

5 free parameters · 5 axioms · 2 invented entities

This is an empirical systems paper. The central claim rests on standard robot-learning modeling choices, a matched-evaluation assumption against DeMaVLA, and several hand-chosen system hyperparameters rather than free physical constants or newly postulated particles.

free parameters (5)
  • video co-training weight λ_vid
    Balances action flow-matching loss against future-video co-training; value is a training hyperparameter that can change the executor representation.
  • action horizon H
    Length of predicted action chunks; chosen for control and RTC, not derived.
  • System 2 history window T=5 frames at 1 Hz
    Planner observation window is fixed by design and affects subtask prediction.
  • planner update interval Δt=2 s
    Deployment coupling rate between planner and executor is hand-chosen.
  • flow-matching noise schedule / denoising setup
    Continuous action generation depends on the chosen flow-matching training and inference schedule.
axioms (5)
  • domain assumption Video co-training at train time yields useful world-aware representations even when future video is not generated at inference.
    Core WAM premise, following Fast-WAM/GigaWorld-style arguments in Sections 2.2–3.3.
  • domain assumption Matching robot platform, pretraining data, post-training data, task protocol, and success criteria isolates policy-class differences between WAM and VLA.
    Load-bearing evaluation assumption for the DeMaVLA folding comparison (Sections 1, 4.1, 4.3).
  • domain assumption Conditional flow matching is an adequate generative model for continuous dual-arm action chunks.
    Action expert formulation in Section 3.3.1.
  • domain assumption A pretrained vision-language annotator can produce reliable subtask boundaries for System 2 supervision.
    Subtask boundary construction in Section 3.2.
  • ad hoc to paper Optional high-level planning should be invoked only when decomposition is useful; atomic tasks should bypass System 2.
    Design principle of DSWAM dual-system control path (Abstract, Section 3.1).
invented entities (2)
  • DSWAM dual-system architecture (System 1 WAM executor + optional System 2 subtask planner) no independent evidence
    purpose: Decouple semantic task decomposition from world-aware physical execution for fine-grained robot manipulation.
    The paper's main proposed system entity; evaluated empirically rather than derived.
  • Transition-aware subtask supervision labels (current subtask / next subtask / done) no independent evidence
    purpose: Train the planner to emit the next executable instruction near boundaries.
    Labeling scheme introduced for System 2 training data construction in Section 3.2.

pith-pipeline@v1.1.0-grok45 · 19254 in / 3347 out tokens · 23345 ms · 2026-07-11T11:20:27.729623+00:00 · methodology

0 comments
read the original abstract

World Action Models (WAMs) provide a promising alternative to Vision-Language-Action (VLA) policies by using video-based world modeling as dense supervision for robot action learning. Existing WAMs excel at physically grounded execution, but typically lack the explicit language-level planning interface in VLM-based VLAs for decomposing coarse instructions. Such decomposition becomes important when household tasks involve complex multi-step goals, where coarse user commands need to be converted into sequences of fine-grained executable subtasks. Meanwhile, the field still lacks a fair real-robot comparison between VLA and WAM execution capabilities, since existing systems often differ in data, robot embodiments, and task protocols. To address both the decomposition gap and the need for a controlled WAM-VLA comparison, we introduce DSWAM, a Dual-System World Action Foundation Model for fine-grained robot manipulation. DSWAM keeps a System 1 WAM executor as the default control path and optionally activates a System 2 vision-language subtask planner only when task decomposition is useful. The planner predicts executable subtasks from short-term visual history and a global task prompt, while the WAM executor performs world-aware action generation for each instruction or subtask. The executor is trained with action prediction and video co-training, but inference directly predicts action chunks without explicit future video generation. To make this execution path practical on real robots, we further integrate TensorRT acceleration, asynchronous execution, and real-time chunking (RTC) so that policy queries do not block robot control. To provide a fair real-robot comparison with VLA policies, we build and evaluate DSWAM under the DeMaVLA real-world deformable manipulation setting with matched robot platform, pretraining data, post-training data, and evaluation criteria.

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

29 extracted references · 25 linked inside Pith

  1. [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, et al. Qwen3-vl technical report.arXiv preprint arXiv:2511.21631, 2025

    Pith/arXiv arXiv 2025

  2. [2]

    Paligemma: A versatile 3b vlm for transfer.arXiv preprint arXiv:2407.07726, 2024

    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

    Pith/arXiv arXiv 2024

  3. [3]

    Motus: Aunifiedlatentactionworldmodel.arXiv preprint arXiv:2512.13030, 2025

    Hongzhe Bi, Hengkai Tan, Shenghao Xie, Zeyuan Wang, Shuhe Huang, Haitian Liu, Ruowen Zhao, Yao Feng, ChendongXiang, YinzeRong, etal. Motus: Aunifiedlatentactionworldmodel.arXiv preprint arXiv:2512.13030, 2025

    Pith/arXiv arXiv 2025

  4. [4]

    Gr00t n1: An open foundation model for generalist humanoid robots

    Johan Bjorck, Fernando Castañeda, Nikita Cherniadev, Xingye Da, Runyu Ding, Linxi Fan, Yu Fang, Dieter Fox, Fengyuan Hu, Spencer Huang, et al. Gr00t n1: An open foundation model for generalist humanoid robots. arXiv preprint arXiv:2503.14734, 2025

    Pith/arXiv arXiv 2025

  5. [5]

    arXiv preprint arXiv:2410.24164, 2024

    Kevin Black, Noah Brown, Danny Driess, Adnan Esmail, Michael Equi, Chelsea Finn, Niccolo Fusai, Lachy Groom, Karol Hausman, Brian Ichter, et al.π0: A vision-language-action flow model for general robot control. arXiv preprint arXiv:2410.24164, 2024

    Pith/arXiv arXiv 2024

  6. [6]

    Real-time execution of action chunking flow policies.Advances in Neural Information Processing Systems, 38:33383–33407, 2026

    Kevin Black, Manuel Galliker, and Sergey Levine. Real-time execution of action chunking flow policies.Advances in Neural Information Processing Systems, 38:33383–33407, 2026

    2026

  7. [7]

    Xiaomi-robotics-0: An open-sourced vision-language-action model with real-time execution.arXiv preprint arXiv:2602.12684, 2026

    Rui Cai, Jun Guo, Xinze He, Piaopiao Jin, Jie Li, Bingxuan Lin, Futeng Liu, Wei Liu, Fei Ma, Kun Ma, et al. Xiaomi-robotics-0: An open-sourced vision-language-action model with real-time execution.arXiv preprint arXiv:2602.12684, 2026

    arXiv 2026

  8. [8]

    Rynnvla-002: A unified vision-language-action and world model.arXiv preprint arXiv:2511.17502, 2025

    Jun Cen, Siteng Huang, Yuqian Yuan, Kehan Li, Hangjie Yuan, Chaohui Yu, Yuming Jiang, Jiayan Guo, Xin Li, Hao Luo, et al. Rynnvla-002: A unified vision-language-action and world model.arXiv preprint arXiv:2511.17502, 2025

    Pith/arXiv arXiv 2025

  9. [9]

    Gr-3 technical report.arXiv preprint arXiv:2507.15493, 2025

    Chilam Cheang, Sijin Chen, Zhongren Cui, Yingdong Hu, Liqun Huang, Tao Kong, Hang Li, Yifeng Li, Yuxiao Liu, Xiao Ma, et al. Gr-3 technical report.arXiv preprint arXiv:2507.15493, 2025

    Pith/arXiv arXiv 2025

  10. [10]

    Robotwin 2.0: A scalable data generator and benchmark with strong domain randomization for robust bimanual robotic manipulation.arXiv preprint arXiv:2506.18088, 2025

    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

    Pith/arXiv arXiv 2025

  11. [11]

    Starvla: A lego-like codebase for vision-language-action model developing.arXiv preprint arXiv:2604.05014, 2026

    StarVLA Community. Starvla: A lego-like codebase for vision-language-action model developing.arXiv preprint arXiv:2604.05014, 2026

    Pith/arXiv arXiv 2026

  12. [12]

    Unified 4d world action modeling from video priors with asynchronous denoising.arXiv preprint arXiv:2604.26694, 2026

    Jun Guo, Qiwei Li, Peiyan Li, Zilong Chen, Nan Sun, Yifei Su, Heyun Wang, Yuan Zhang, Xinghang Li, and Huaping Liu. Unified 4d world action modeling from video priors with asynchronous denoising.arXiv preprint arXiv:2604.26694, 2026

    Pith/arXiv arXiv 2026

  13. [13]

    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

    Pith/arXiv arXiv 2025

  14. [14]

    Physical Intelligence, Kevin Black, Noah Brown, James Darpinian, Karan Dhabalia, Danny Driess, Adnan Es- mail, Michael Equi, Chelsea Finn, Niccolo Fusai, et al.π0.5: a vision-language-action model with open-world generalization.arXiv preprint arXiv:2504.16054, 2025

    Pith/arXiv arXiv 2025

  15. [15]

    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

    Pith/arXiv arXiv 2025

  16. [16]

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

    Moo Jin Kim, Karl Pertsch, Siddharth Karamcheti, Ted Xiao, Ashwin Balakrishna, Suraj Nair, Rafael Rafailov, Ethan Foster, Grace Lam, Pannag Sanketi, et al. Openvla: An open-source vision-language-action model.arXiv preprint arXiv:2406.09246, 2024

    Pith/arXiv arXiv 2024

  17. [17]

    Cosmos policy: Fine-tuning video models for visuomotor control and planning.arXiv preprint arXiv:2601.16163, 2026

    Moo Jin Kim, Yihuai Gao, Tsung-Yi Lin, Yen-Chen Lin, Yunhao Ge, Grace Lam, Percy Liang, Shuran Song, Ming-Yu Liu, Chelsea Finn, and Jinwei Gu. Cosmos policy: Fine-tuning video models for visuomotor control and planning.arXiv preprint arXiv:2601.16163, 2026. 12

    Pith/arXiv arXiv 2026

  18. [18]

    Causal world modeling for robot control.arXiv preprint arXiv:2601.21998, 2026

    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

    Pith/arXiv arXiv 2026

  19. [19]

    Flow matching for generative modeling.arXiv preprint arXiv:2210.02747, 2022

    Yaron Lipman, Ricky TQ Chen, Heli Ben-Hamu, Maximilian Nickel, and Matt Le. Flow matching for generative modeling.arXiv preprint arXiv:2210.02747, 2022

    Pith/arXiv arXiv 2022

  20. [20]

    Motubrain: An advanced world action model for robot control.arXiv preprint arXiv:2604.27792, 2026

    MotuBrain Team, Chendong Xiang, Fan Bao, Haitian Liu, Hengkai Tan, Hongzhe Bi, James Li, Jiabao Liu, Jingrui Pang, Kiro Jing, et al. Motubrain: An advanced world action model for robot control.arXiv preprint arXiv:2604.27792, 2026

    Pith/arXiv arXiv 2026

  21. [21]

    Fast: Efficient action tokenization for vision-language-action models.arXiv preprint arXiv:2501.09747, 2025

    Karl Pertsch, Kyle Stachowicz, Brian Ichter, Danny Driess, Suraj Nair, Quan Vuong, Oier Mees, Chelsea Finn, and Sergey Levine. Fast: Efficient action tokenization for vision-language-action models.arXiv preprint arXiv:2501.09747, 2025

    Pith/arXiv arXiv 2025

  22. [22]

    DeMaVLA: A vision-language-action foundation model for generalizable deformable manipulation.arXiv preprint arXiv:2605.31286, 2026

    Taiyi Su, Jian Zhu, Tianjian Wang, Youzhang He, Zitai Huang, Jianjun Zhang, Chong Ma, Hanyang Wang, Tianjiao Zhang, Munan Yin, Weihao Ding, and Yi Xu. DeMaVLA: A vision-language-action foundation model for generalizable deformable manipulation.arXiv preprint arXiv:2605.31286, 2026

    Pith/arXiv arXiv 2026

  23. [23]

    A pragmatic vla foundation model.arXiv preprint arXiv:2601.18692, 2026

    Wei Wu, Fan Lu, Yunnan Wang, Shuai Yang, Shi Liu, Fangjing Wang, Qian Zhu, He Sun, Yong Wang, Shuailei Ma, et al. A pragmatic vla foundation model.arXiv preprint arXiv:2601.18692, 2026

    Pith/arXiv arXiv 2026

  24. [24]

    Abot-m0: Vla foundation model for robotic manipulation with action manifold learning.arXiv preprint arXiv:2602.11236, 2026

    Yandan Yang, Shuang Zeng, Tong Lin, Xinyuan Chang, Dekang Qi, Junjin Xiao, Haoyun Liu, Ronghan Chen, Yuzhi Chen, Dongjie Huo, et al. Abot-m0: Vla foundation model for robotic manipulation with action manifold learning.arXiv preprint arXiv:2602.11236, 2026

    Pith/arXiv arXiv 2026

  25. [25]

    Gigaworld-policy: An efficient action-centered world–action model.arXiv preprint arXiv:2603.17240, 2026

    Angen Ye, Boyuan Wang, Chaojun Ni, Guan Huang, Guosheng Zhao, Hao Li, Hengtao Li, Jie Li, Jindi Lv, Jingyu Liu, et al. Gigaworld-policy: An efficient action-centered world–action model.arXiv preprint arXiv:2603.17240, 2026

    arXiv 2026

  26. [26]

    World action models are zero-shot policies.arXiv preprint arXiv:2602.15922, 2026

    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

    Pith/arXiv arXiv 2026

  27. [27]

    Fast-WAM: Do world action models need test-time future imagination?arXiv preprint arXiv:2603.16666, 2026

    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

    Pith/arXiv arXiv 2026

  28. [28]

    X-vla: Soft-prompted transformer as scalable cross-embodiment vision-language- action model.arXiv preprint arXiv:2510.10274, 2025

    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

    Pith/arXiv arXiv 2025

  29. [29]

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

    Brianna Zitkovich, Tianhe Yu, Sichun Xu, Peng Xu, Ted Xiao, Fei Xia, Jialin Wu, Paul Wohlhart, Stefan Welker, Ayzaan 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. 13

    2023