Writing study: Structural Action Transformer for 3D Dexterous Manipulation

CVPR overviewCorpus atlasBy contribution typeVisual galleryRules & templateWriting patternsPaper studies

← CVPR 2026 oral writing corpus

Xiaohan Lei, Min Wang, Bohong Weng, Wengang Zhou, Houqiang Li · Official CVPR 2026 oral record

Observed writing choices, not acceptance rules. This study examines the inspected manuscript version. It does not establish why the paper was accepted or that its presentation choices should be copied unchanged.

Figure, table and section ledger

Editorial contribution type: Architecture & representation (primary).

Layout extraction, not a required paper template. Counts are distinct caption labels; a multi-panel figure counts once. Section spans are estimated from heading positions, including floats and whitespace. Method and limitations.

Canonical PDF

12 physical pages

Before references / appendix

≈ 8.05 page-equivalents

Main figures / tables

8 / 6

Attached appendix figures / tables

0 / 0

Attached appendix span

Separate file recorded below; not in this PDF total

Before references / appendix: ≈ 67.08%References / intervening material: ≈ 32.92%Estimated layout envelopes, not text/artwork area. Unresolved appendix boundaries remain within the undivided post-body material.

Extraction notes: No attached appendix boundary detected. Zero means no recognized caption, not independently verified absence.

Section-by-section lengths and source pages

The title/abstract/front matter precede the first numbered heading. A span ends at the next detected top-level heading, not at the last sentence; end-page whitespace is included. References and statements are listed separately. These are not measurements of text-only or artwork area.

Top-level heading	Starts on PDF page	Estimated span	Share of entire PDF
1. Introduction	1	≈ 1.25 pp.	10.42%
2. Related Work	2	≈ 0.95 pp.	7.92%
3. Method	3	≈ 1.65 pp.	13.75%
4. Experiments	5	≈ 3.18 pp.	26.5%
5. Conclusion	8	≈ 0.46 pp.	3.83%
6. Acknowledgments	8	≈ 0.16 pp.	1.33%
References	9	≈ 3.95 pp.	32.92%

All detected figure and table locations

Main text

• Figure 1 · p. 1
• Figure 2 · p. 4
• Figure 3 · p. 5
• Table 1 · p. 5
• Table 2 · p. 6
• Table 3 · p. 6
• Figure 6 · p. 6
• Table 4 · p. 7
• Figure 4 · p. 7
• Figure 5 · p. 7
• Table 5 · p. 7
• Figure 7 · p. 8
• Figure 8 · p. 8
• Table 6 · p. 8

Attached appendix

No recognized caption labels.

Separate source files—excluded from canonical-PDF totals

A separate author version may repeat main text and references. Whole-file lengths and counts are not extra appendix material and are never added to the canonical PDF totals. Read the study's source scope for which sections received substantive reading versus layout inspection.

Additional source 1

Separate PDF (or ZIP containing it): whole-file layout of 6 physical pages; 2 detected figures and 3 tables. These may include overlapping main-text material.

Located appendix only: starts on physical p.1; ≈ 5.94 page-equivalents, 2 figure labels and 3 table labels. This interval is distinct from the whole file.

Detected heading	PDF page	Estimated span
Supplementary Material	1	≈ 0.02 pp.
1. Embodied Joint Codebook Mapping Table	1	≈ 0.07 pp.
2. 3D Re-implementation of HPT and UniAct	1	≈ 0.59 pp.
3. Visualization of Failure Cases	1	≈ 0.12 pp.
4. Ablation on Action Horizon Length	1	≈ 0.85 pp.
5. Real-World Demonstration Video	2	≈ 0.1 pp.
References	2	≈ 4.19 pp.

• Figure 1 · p. 1
• Table 1 · p. 2
• Figure 2 · p. 2
• Table 2 · p. 2
• Table 3 · p. 3

Argument and organization

1. Change the token axis

   Contrast temporal and structural action views before defining the conditional flow policy.

2. Architecture and transfer evidence

   Describe observation/action tokenizers, then separate simulation, data ablations, few-shot adaptation and real-world tasks.

Problem, claim and contribution

The opening illustrates the actual axis swap in the action tensor, making cross-embodiment handling a representation question rather than a generic scaling claim.

Figure 1 and Introduction · Main paper · Additional source 2

Argument and section order

Problem formulation and flow parameterization precede the network, while experiments separate pretraining mixture effects from individual architectural components.

Sections 3.1–3.3 and 4 · Main paper · Additional source 2

Figures, captions and visual explanation

Figure 2 labels point-cloud, language and joint-codebook streams, shows the transpose/projection on actions and ends with the ODE integration step.

Figure 2, PDF p. 4 · Main paper · Additional source 2

Tables, comparisons and evidence

Component/codebook ablations share a success metric, while the few-shot plot compares training curves at several demonstration budgets instead of one final point.

Tables 4–5 and Figure 4, PDF p. 7 · Main paper · Additional source 2

Appendix and reproducibility

A long codebook mapping table makes cross-embodiment correspondences explicit; failure analysis names missing haptic/force information instead of attributing every error to insufficient data.

Supplement §§1–4; Table 3 pp. 3–6 · Main paper · Additional source 2

Clarity and writing conventions

The failure discussion names occlusion and kinematic/contact mismatch, tying limitations to the information actually available to the policy.

Section 4.5, PDF p. 8 · Main paper · Additional source 2

What to reuse—and what not to infer

Transferable lesson: When novelty is a tensor reorganization, draw the axes and operation explicitly, then separate representation ablations from extra-data benefits.

Boundary: A single-camera failure and embodiment mismatch remain despite cross-embodiment pretraining; the selected real-world tasks are not universal dexterous competence.

Apply this lesson using the argument and evidence worksheet.

Inspection record and sources

Checked

2026-09-26

Reading status

Writing-focused manuscript scan + selected visuals inspected

Main text

Manuscript-wide writing scan of main PDF pp. 1–8: framing, section structure, method exposition, figure/table captions, evaluation organization and conclusions; not an exhaustive technical proof audit or reproduction.

Appendix / supplement

Scanned the six-page official ZIP-member PDF for baseline reimplementations, force/occlusion failures, horizon tests, rollout stills, and the full embodied-joint codebook mapping; demonstration videos were not viewed.

Visually inspected PDF pages

4, 7

This is a writing-focused scan of the manuscript and the stated supplementary scope: framing, organization, visual explanation, evidence presentation and wording. It is not a line-by-line proof audit, reproduction of results or exhaustive scientific review.

• Official CVPR 2026 CVF proceedings PDF; retrieved 2026-09-26 · 12 pages
• Official CVPR 2026 supplementary PDF extracted from CVF ZIP archive member; retrieved 2026-09-26 · 6 pages