VLA models through the lens of adaptation

\[\newcommand{\obs}{o} \newcommand{\act}{a} \newcommand{\policy}{\pi} \newcommand{\R}{\mathbb{R}}\]

This is a survey-first note: map the field and its sub-problems before any one system takes the stage. Qwen-VLA (Qwen Team, 2026) appears only as one point in a design space that exists with or without it. Numbers attributed to a paper are stated as that paper reports them; my own judgment is quarantined to §D.

A. The field first

A.1 What a vision-language-action model is

A vision-language-action (VLA) model is a policy that maps visual observations $\obs_t$ and a natural-language instruction $\ell$ to robot actions $\act_t$:

\[\policy_\theta:\; (\obs_{1:t},\, \ell)\; \longmapsto\; \act_t \in \mathcal{A}.\]

The lineage is short and fast. RT-1 (Brohan et al., 2023) showed a single transformer absorbing large-scale real-robot data; RT-2 (Brohan et al., 2023) made the decisive move of reusing a vision-language model’s web knowledge by emitting actions as text tokens; PaLM-E (Driess et al., 2023) folded continuous sensor states into the language model itself. OpenVLA (Kim et al., 2024) and Octo (Octo Model Team et al., 2024) opened the recipe; RoboFlamingo (Li et al., 2024) showed a VLM backbone is an effective imitator; $\pi_0$ (Black et al., 2024) replaced discrete action tokens with a flow-matching continuous decoder.

A.2 Why adaptation is the right axis

It is tempting to rank VLAs by a headline success rate. That is the wrong axis. A policy that scores 97% on one benchmark suite and collapses on a new robot, a new task family, or a mildly shifted environment has not demonstrated a foundation model — it has demonstrated overfitting at scale.

Claim (the axis). The quantity that separates a benchmark policy from an embodied foundation model is adaptation: how cheaply one set of weights survives a change of body, task, action space, or distribution.

The rest of this note is organized around those four changes.

B. The four adaptations

B.1 Embodiment adaptation — one policy, many bodies

Different robots have different kinematics, sensors, and control conventions. The naive fix is a per-robot output head; it does not scale and it forbids transfer. Two lines attack this. Open X-Embodiment (Open X-Embodiment Collaboration et al., 2024) pooled 22 robots into one dataset and trained RT-X models that transfer across bodies; CrossFormer (Doshi et al., 2024) pushed further — a single transformer across manipulation, navigation, locomotion, and aviation with no manual action/observation-space alignment. On the navigation side, ViNT (Shah et al., 2023) is a cross-robot navigation foundation model with positive transfer to unseen platforms.

The newest lever is to describe the body in language. Instead of a learned embodiment embedding, a textual description of the current robot and its control convention is placed in the prompt — so the same weights specialize at inference by reading who they are driving. This is the “tool-schema” pattern from software agents, imported into robotics.

B.2 Task adaptation — manipulation, navigation, and trajectory in one model

Manipulation, navigation, and trajectory prediction are historically three fields with three benchmark cultures. The generalist bet is that they are one problem — conditional sequence generation over actions — and that joint training transfers. GATO (Reed et al., 2022) made the maximalist version of this bet (one network, 600+ tasks across modalities). In robotics specifically, the unification is now concrete: navigation is cast as action-and-trajectory prediction over the same interface as manipulation, evaluated on VLN benchmarks R2R (Anderson et al., 2018) and RxR (Ku et al., 2020) alongside manipulation suites.

B.3 Action-space adaptation — discrete tokens vs continuous flow

How a VLA emits an action is a real design axis, not an implementation detail.

Flow matching, in one line. Learn a velocity field $v_\theta(x,\tau\mid c)$ so that integrating the ODE $\;\dot{x}=v_\theta(x,\tau\mid c)\;$ from noise $x_0\sim \mathcal{N}(0,I)$ at $\tau{=}0$ to $\tau{=}1$ lands on a valid action chunk $x_1$, conditioned on the VLM’s language+vision tokens $c$. Continuous, high-precision, and it sidesteps action quantization — at the cost of an ODE solve per action.

B.4 Distribution / OOD adaptation — the honest frontier

The final and hardest change is distribution. SimplerEnv (Li et al., 2024) exists precisely to measure the sim-to-real and cross-setup gap for manipulation; LIBERO (Liu et al., 2023) probes lifelong transfer across task suites; real-world ALOHA (Zhao et al., 2023) out-of-distribution trials and dynamic manipulation (moving targets) are where reported numbers fall hardest. This is the axis on which “foundation model” claims should be stress-tested, not the in-distribution suites.

C. Qwen-VLA as the worked case

Qwen-VLA (Qwen Team, 2026) threads all four axes:

  • Action space — a 1.15B DiT flow-matching action decoder on a Qwen3.5-4B VLM backbone (continuous, §B.3).
  • Embodiment — embodiment-aware prompt conditioning: the body and control convention are described in text, no per-platform head (§B.1).
  • Task — manipulation, navigation, and trajectory folded into one action-and-trajectory objective (§B.2).
  • Distribution — evaluated out to real-world ALOHA OOD and dynamic manipulation (§B.4).

Reported scores (as stated by the paper): 97.9% LIBERO, 73.7% Simpler-WidowX, 86.1/87.2% RoboTwin (Mu et al., 2025) Easy/Hard, 69.0% R2R OSR, 59.6% RxR SR, 76.9% avg ALOHA OOD, and 26.6% zero-shot on DOMINO (Fang et al., 2026) dynamic manipulation. The full reading note is here.

D. Reading it critically

(This section is deliberately cautious — hedged where the evidence is.)

A benchmark table is a claim, not a proof. Five questions decide whether the numbers mean progress:

  1. Baselines. Is each score against a specialist SOTA or a weak generalist? A high number against the wrong baseline measures nothing.
  2. Train–test split. How much of LIBERO/R2R/RxR distribution entered the joint pretraining mix? Generalist scores are only credible if the evaluation split was genuinely held out.
  3. Embodiment-OOD definition. “New embodiment” is meaningful only if the control convention was truly unseen — otherwise it is in-distribution wearing a costume.
  4. Action normalization. Cross-embodiment results hinge on how joint/action spaces are normalized; a favorable normalization can manufacture transfer.
  5. Inference latency. A flow-matching ODE solve is not free. Real-robot control rate — not just success rate — determines whether the policy is deployable.

The number that reads both ways. DOMINO zero-shot 26.6% is simultaneously evidence that cross-task transfer to dynamic manipulation is non-trivially possible and evidence that dynamic manipulation remains largely unsolved. Promise and open problem in one figure — and a caution against reading the in-distribution 97.9% as the headline.

E. Where it points

The trajectory is clear even if the destination is not. The field is converging on: continuous flow/diffusion decoders over discrete tokens; language-described embodiment over learned embodiment embeddings; and joint manipulation-navigation- trajectory training over siloed policies. What remains genuinely open — and where I would place research effort — is the fourth axis: dynamic, out-of-distribution adaptation, where even the strongest current numbers are low. A VLA that is an embodied foundation model, rather than a very good benchmark policy, will be decided there.

Reference map

Not a flat list — a clustered web. Each paper links to its [arXiv] (formal) and, where one exists, an [on this blog] deep-note (informal). The → edges are the spider: follow them to jump from any paper to the work it builds on or leads to.

① The VLA lineage — scale → web-knowledge → open → continuous

RT-1 proved scale on real-robot data; RT-2 turned actions into web-knowledge text tokens; OpenVLA / Octo opened the recipe; π0 made the decoder continuous — the move that Qwen-VLA inherits.

② How actions are emitted — discrete tokens vs continuous flow

The fork of §B.3: quantize actions into tokens, or generate them continuously.

  • FAST · arXiv — frequency-space action tokens. → discrete counterpart to Diffusion Policy; pairs with RT-2
  • Diffusion Policy · arXiv · project — action chunk as denoising. → architecture DiT; chunking ACT
  • ACT / ALOHA · arXiv · project — action chunking + low-cost bimanual. → also a benchmark, see ④
  • DiT · arXiv — transformer diffusion backbone. → the decoder in π0 and Qwen-VLA
  • Flow Matching · arXiv — the training objective. → used by π0, Qwen-VLA

③ One policy, many bodies — embodiment & task unification

The §B.1–B.2 bet: pool bodies and tasks into one policy.

  • Open X-Embodiment · arXiv · project — 22-robot pooled dataset + RT-X. → scaled by CrossFormer
  • CrossFormer · arXiv · project — one transformer across 20 embodiments. → language-described bodies in Qwen-VLA
  • ViNT · arXiv · project — cross-robot navigation foundation. → navigation benchmarks R2R/RxR
  • NaVILA (Cheng et al., 2024) · arXiv — VLA for legged-robot navigation; mid-level language actions bridge to a low-level locomotion policy. → sibling unification attempt to Uni-NaVid
  • Uni-NaVid (Zhang et al., 2024) · arXiv — one video-based VLA unifying four navigation sub-tasks (instruction-following, object search, QA, tracking) instead of one model per task.
  • GATO · arXiv · blog — one net, 600+ tasks. → maximalist ancestor of Qwen-VLA
  • RoboCat (Bousmalis et al., 2023) · arXiv · blog — self-improving generalist: few-shot adapt to a new embodiment, then generate more of its own training data. → same multi-embodiment bet as CrossFormer

④ Where we measure — benchmarks

The §B.4 proving grounds. Qwen-VLA is scored on all of these.

  • LIBERO · arXiv · project — lifelong-transfer manipulation.
  • SimplerEnv · arXiv · project — real-to-sim manipulation eval.
  • R2R · arXiv — vision-and-language navigation. → multilingual scale-up RxR
  • RxR · arXiv — multilingual, denser VLN.
  • ALOHA (real-world OOD) · arXiv · project — the OOD stress test. → method side ②

⑤ The anchor

  • Qwen-VLA · arXiv · on this blog — pulls the flow-matching decoder from ②, language-described embodiment from ③, and is scored on ④. The worked case of §C.
Formal cited bibliography (auto-generated)

2026

  1. Qwen-VLA: Unifying Vision-Language-Action Modeling across Tasks, Environments, and Robot Embodiments
    Qwen Team
    arXiv preprint arXiv:2605.30280, 2026
  2. Towards Generalizable Robotic Manipulation in Dynamic Environments
    Heng Fang, Shangru Li, Shuhan Wang, and 3 more authors
    In European Conference on Computer Vision (ECCV), 2026
    DOMINO; verify fields before relying on them

2025

  1. FAST: Efficient Action Tokenization for Vision-Language-Action Models
    Karl Pertsch, Kyle Stachowicz, Brian Ichter, and 6 more authors
    arXiv preprint arXiv:2501.09747, 2025
  2. RoboTwin: Dual-Arm Robot Benchmark with Generative Digital Twins
    Yao Mu, Tianxing Chen, Zanxin Chen, and 11 more authors
    arXiv preprint, 2025
    Verify fields before relying on them

2024

  1. OpenVLA: An Open-Source Vision-Language-Action Model
    Moo Jin Kim, Karl Pertsch, Siddharth Karamcheti, and 2 more authors
    In Conference on Robot Learning (CoRL), 2024
    arXiv:2406.09246
  2. Octo: An Open-Source Generalist Robot Policy
    Octo Model Team, Dibya Ghosh, Homer Walke, and 2 more authors
    In Robotics: Science and Systems (RSS), 2024
    arXiv:2405.12213
  3. Vision-Language Foundation Models as Effective Robot Imitators
    Xinghang Li, Minghuan Liu, Hanbo Zhang, and 3 more authors
    In International Conference on Learning Representations (ICLR), 2024
    arXiv:2311.01378
  4. \pi_0: A Vision-Language-Action Flow Model for General Robot Control
    Kevin Black, Noah Brown, Danny Driess, and 3 more authors
    arXiv preprint arXiv:2410.24164, 2024
    Physical Intelligence technical report
  5. Open X-Embodiment: Robotic Learning Datasets and RT-X Models
    Open X-Embodiment Collaboration, Abhishek Padalkar, and others
    In IEEE International Conference on Robotics and Automation (ICRA), 2024
    arXiv:2310.08864
  6. Scaling Cross-Embodied Learning: One Policy for Manipulation, Navigation, Locomotion and Aviation
    Ria Doshi, Homer Walke, Oier Mees, and 2 more authors
    In Conference on Robot Learning (CoRL), 2024
    arXiv:2408.11812
  7. Evaluating Real-World Robot Manipulation Policies in Simulation
    Xuanlin Li, Kyle Hsu, Jiayuan Gu, and 3 more authors
    In Conference on Robot Learning (CoRL), 2024
    SimplerEnv; arXiv:2405.05941
  8. NaVILA: Legged Robot Vision-Language-Action Model for Navigation
    An-Chieh Cheng, Yandong Ji, Zhaojing Yang, and 7 more authors
    arXiv preprint, 2024
    Verify fields before relying on them
  9. Uni-NaVid: A Video-based Vision-Language-Action Model for Unifying Embodied Navigation Tasks
    Jiazhao Zhang, Kunyu Wang, Shaoan Wang, and 6 more authors
    arXiv preprint, 2024
    Verify fields before relying on them

2023

  1. RT-1: Robotics Transformer for Real-World Control at Scale
    Anthony Brohan, Noah Brown, Justice Carbajal, and 1 more author
    In Robotics: Science and Systems (RSS), 2023
    arXiv:2212.06817
  2. RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Control
    Anthony Brohan, Noah Brown, Danny Driess, and 2 more authors
    In Conference on Robot Learning (CoRL), 2023
    arXiv:2307.15818; proceedings lead author Zitkovich et al.
  3. PaLM-E: An Embodied Multimodal Language Model
    Danny Driess, Fei Xia, Mehdi S. M. Sajjadi, and 2 more authors
    In International Conference on Machine Learning (ICML), 2023
    arXiv:2303.03378
  4. ViNT: A Foundation Model for Visual Navigation
    Dhruv Shah, Ajay Sridhar, Nitish Dashora, and 4 more authors
    In Conference on Robot Learning (CoRL), 2023
    arXiv:2306.14846
  5. Diffusion Policy: Visuomotor Policy Learning via Action Diffusion
    Cheng Chi, Zhenjia Xu, Siyuan Feng, and 5 more authors
    In Robotics: Science and Systems (RSS), 2023
    arXiv:2303.04137
  6. Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardware
    Tony Z. Zhao, Vikash Kumar, Sergey Levine, and 1 more author
    In Robotics: Science and Systems (RSS), 2023
    ALOHA/ACT; arXiv:2304.13705
  7. Scalable Diffusion Models with Transformers
    William Peebles and Saining Xie
    In International Conference on Computer Vision (ICCV), 2023
    arXiv:2212.09748
  8. Flow Matching for Generative Modeling
    Yaron Lipman, Ricky T. Q. Chen, Heli Ben-Hamu, and 2 more authors
    In International Conference on Learning Representations (ICLR), 2023
    arXiv:2210.02747
  9. LIBERO: Benchmarking Knowledge Transfer for Lifelong Robot Learning
    Bo Liu, Yifeng Zhu, Chongkai Gao, and 4 more authors
    In NeurIPS Datasets and Benchmarks Track, 2023
    arXiv:2306.03310
  10. RoboCat: A Self-Improving Generalist Agent for Robotic Manipulation
    Konstantinos Bousmalis, Giulia Vezzani, Dushyant Rao, and 8 more authors
    arXiv preprint, 2023
    Verify fields before relying on them

2022

  1. A Generalist Agent
    Scott Reed, Konrad Zolna, Emilio Parisotto, and 1 more author
    Transactions on Machine Learning Research (TMLR), 2022
    arXiv:2205.06175

2020

  1. Room-Across-Room: Multilingual Vision-and-Language Navigation with Dense Spatiotemporal Grounding
    Alexander Ku, Peter Anderson, Roma Patel, and 2 more authors
    In Empirical Methods in Natural Language Processing (EMNLP), 2020
    RxR; arXiv:2010.07954

2018

  1. Vision-and-Language Navigation: Interpreting Visually-Grounded Navigation Instructions in Real Environments
    Peter Anderson, Qi Wu, Damien Teney, and 6 more authors
    In IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2018
    R2R; arXiv:1711.07280

All entries are established works with verified arXiv identifiers, including RoboTwin, DOMINO, NaVILA, Uni-NaVid, and RoboCat — added 2026-09-20 (previously named in prose but not formally cited). Informal / project-page links can be layered onto each node next (verified before adding — none fabricated).