Writing study: HABIT: Human-Aware Behavior and Interaction Training Dataset for Robot Manipulation

NeurIPS overviewCorpus atlasBy contribution typeVisual galleryRules & templateWriting patternsPaper studies

← NeurIPS 2026 oral writing corpus

Jaehwi Song, Suchae Jeong, Byeongguk Jeon, Sungdong Kim, Hyungmok Son, Kimin Lee · Official NeurIPS 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: Datasets, benchmarks & measurement (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

30 physical pages

Before references / appendix

≈ 10.09 page-equivalents

Main figures / tables

9 / 0

Attached appendix figures / tables

17 / 3

Attached appendix span

≈ 16.91 page-equivalents

Before references / appendix: ≈ 33.63%References / intervening material: ≈ 10%Attached appendix: ≈ 56.37%Estimated layout envelopes, not text/artwork area. Unresolved appendix boundaries remain within the undivided post-body material.

Source-bound correction: Top-level heading and reference/appendix boundaries checked against source PDF text. Bibliography titles/author initials, numbered contribution/property bullets, diagram steps, equations and TOC entries are not substantive headings. Section spans are page-coordinate estimates, not ink area or prose-only length. This is an author manuscript mirror, not a verified accepted-version template. Source layout uses physical single-column y/page-height, including equation-heavy pages.

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	≈ 1.17 pp.	3.9%
2 HABIT Dataset	2	≈ 3.38 pp.	11.27%
3 Evaluation Framework	6	≈ 0.9 pp.	3%
4 Experiments	7	≈ 3 pp.	10%
5 Related Work	10	≈ 0.43 pp.	1.43%
6 Limitations	10	≈ 0.15 pp.	0.5%
7 Conclusion and Future Directions	10	≈ 0.42 pp.	1.4%
References	11	≈ 3 pp.	10%
Appendix:	14	≈ 0.39 pp.	1.3%
A Hardware and Collection Details	14	≈ 0.69 pp.	2.3%
B Evaluation Task Details	15	≈ 8.23 pp.	27.43%
C Model Training Details	23	≈ 1.42 pp.	4.73%
D Main Experiment Details	24	≈ 0.97 pp.	3.23%
E OOD Robustness Analysis	25	≈ 0.96 pp.	3.2%
F Mid-Training Experiment Details	26	≈ 1.87 pp.	6.23%
G Ethics Statement	28	≈ 2.38 pp.	7.93%

All detected figure and table locations

Main text

• Figure 1 · p. 2
• Figure 2 · p. 3
• Figure 3 · p. 4
• Figure 4 · p. 5
• Figure 5 · p. 6
• Figure 6 · p. 8
• Figure 7 · p. 8
• Figure 8 · p. 9
• Figure 9 · p. 9

Attached appendix

• Figure 10 · p. 14
• Figure 11 · p. 16
• Figure 12 · p. 17
• Figure 13 · p. 17
• Figure 14 · p. 18
• Figure 15 · p. 18
• Figure 16 · p. 19
• Figure 17 · p. 20
• Figure 18 · p. 21
• Figure 19 · p. 21
• Figure 20 · p. 22
• Figure 21 · p. 22
• Figure 22 · p. 23
• Table 1 · p. 24
• Table 2 · p. 24
• Table 3 · p. 25
• Figure 23 · p. 26
• Figure 24 · p. 27
• Figure 25 · p. 28
• Figure 26 · p. 29

Argument and organization

1. Problem and contribution

   Introduces human presence as a missing axis of demonstration diversity, then turns it into three operational roles rather than leaving human awareness as an informal slogan.

2. Argument sequence

   Dataset design, collection and statistics appear before the evaluation framework; experiments distinguish task success, role-specific failures and adaptation to new tasks.

3. Evidence and verification

   Each cell's denominator is twenty trials, and the caption explains missing comparisons. Dedicated failure plots and the OOD appendix address whether better aggregate success corresponds to the intended behavior.

Problem, claim and contribution

Introduces human presence as a missing axis of demonstration diversity, then turns it into three operational roles rather than leaving human awareness as an informal slogan.

Introduction and Section 2, pp. 1–3 · Main paper

Argument and section order

Dataset design, collection and statistics appear before the evaluation framework; experiments distinguish task success, role-specific failures and adaptation to new tasks.

Sections 2–4, pp. 2–9 · Main paper

Figures, captions and visual explanation

Figure 6 repeats the same task order for two backbones and explicitly marks unavailable robot-only collaborator comparisons as N/A. Figure 7 overlays desired and violating motions on concrete failure examples. Figure 3 pairs a photographed acquisition arrangement with a two-ring role distribution: inner task counts and outer episode counts are explicitly different units. Camera locations are marked on the physical setup. The footnote distinguishes the five recorded/released views from the robot-side views used for the reported policy training.

Figures 6–7, p. 8 (render inspected); Figure 3 and camera-use footnote, p. 4 (additional rendered inspection) · Main paper

Tables, comparisons and evidence

Each cell's denominator is twenty trials, and the caption explains missing comparisons. Dedicated failure plots and the OOD appendix address whether better aggregate success corresponds to the intended behavior.

Section 4.1 and Figure 6, pp. 7–8; Appendix E, pp. 25–28 · Main paper

Appendix and reproducibility

Task-specific layouts and human/robot action sequences make the benchmark repeatable; training hyperparameters and per-task episode counts are separate tables.

Appendices A–D, pp. 14–25 · Main paper

Clarity and writing conventions

Role names recur consistently in taxonomy, collection, evaluation and limitations; a dedicated limitations section keeps embodiment and human-diversity constraints visible.

Sections 2, 3 and 6, pp. 2–7 and 10 · Main paper

What to reuse—and what not to infer

Transferable lesson: A dataset story should carry its taxonomy through collection, metrics and failure analysis, with explicit trial counts and inapplicable baseline cells.

Boundary: The shown bars do not include uncertainty intervals. Do not treat a twenty-trial mean or a selected six-task evaluation as a precise estimate across all sixty collection tasks.

Apply this lesson using the argument and evidence worksheet.

Inspection record and sources

Checked

2026-10-02

Reading status

Writing-focused manuscript scan + selected visuals inspected

Main text

Writing-oriented scan across manuscript framing, section sequence, figure/table captions, selected argument/evaluation prose and conclusion; not exhaustive technical or proof verification. Sections 1–7, pp. 1–10; references pp. 11–13 Source is the exact public author version listed below, not a verified accepted/camera-ready revision.

Appendix / supplement

Writing-oriented appendix/documentation scan: Appendices A–F, pp. 14–30: cameras/coordinates, task randomization and workflow, training configurations, failure categories, OOD and sample-efficiency details. Source code, data, model outputs and external video assets were not executed or independently validated.

Visually inspected PDF pages

4, 8

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.

• Public author arXiv manuscript 2606.31682v1; not verified identical to the accepted or camera-ready OpenReview revision · 30 pages