ICRA Submission
Final-gate audit for a direct ICRA conference submission on PaperPlaza
(ras.papercept.net), the RAS submission system used across ICRA, IROS, and RA-L.
Everything here is anchored to the verified ICRA 2026 cycle; the numbers move, so
open the current year's "call for papers" page before trusting any of them.
Hard limits (2026-cycle values — reverify each year)
| Check |
2026 rule |
Consequence if violated |
| Page count at submission |
6 pages of content + up to 2 pages containing only references |
Returned without review |
| Format |
IEEE two-column conference template (RAS PaperPlaza support page templates), PDF only |
Compliance failure at upload |
| Anonymity |
Double-anonymous: no authors/affiliations in the PDF; full author list entered in PaperPlaza metadata |
Administrative rejection |
| Video attachment |
≤ 180 seconds, ≤ 20 MB, one of mpeg/mp4/mpg |
Rejected by the upload form |
| Video timing |
Only inside the posted windows (2026: Aug 5–Sep 9 and Sep 17–22) |
No video on record, no late path |
| Deadline |
Sep 15, 2025, 23:59 Pacific (not AoE) |
Closed portal |
The Pacific-time detail is a live trap: authors calibrated to AoE deadlines at ML
conferences lose up to five hours.
The 2026 anonymity turn
Through ICRA 2025, review was single-blind — author names printed on page one.
The 2026 cycle switched to double-anonymous. Concretely:
- Strip the author block, affiliations, and identifying acknowledgements/grant
numbers from the PDF; keep the full, final author list in PaperPlaza metadata
(it cannot be extended casually later — treat the metadata list as binding).
- Rewrite platform tells: "our previously developed XYZ-Arm [12]" becomes "the
XYZ-Arm [12]" with third-person citation.
- Blur or crop lab-identifying scenery in figures and in the video attachment —
posters on the wall, safety vests with logos, institute door signs.
- Anonymize linked material: a GitHub URL with a lab organization name is an
identity leak. Use an anonymized mirror or omit the link at submission.
- Because the policy is one cycle old, confirm it still holds for the target year
before de-anonymizing or re-anonymizing anything.
PaperPlaza mechanics worth knowing in advance
- Every author needs a PaperPlaza PIN; collecting PINs from six co-authors on
deadline night is a known time sink — do it the week the site opens.
- The system runs its own PDF compliance test (fonts embedded, page size, page
count); a failing PDF can be re-uploaded until the deadline, so test early.
- Keywords/session topics are chosen from a fixed RAS taxonomy and steer which
Associate Editor sees the paper — pick the two or three closest, not the
broadest.
- The submission confirmation email is the only receipt; archive it.
PDF craft the compliance checker enforces
- Fonts must be embedded — the usual failures are figures exported from
MATLAB/matplotlib with Type 3 or missing fonts; regenerate with embedded
TrueType/Type 1 before deadline week.
- US Letter page size per the IEEE template; a European A4 build fails quietly.
- Vector figures where possible: reviewers zoom into plots of contact forces
and trajectories, and rasterized 300-dpi exports turn to mush at 400%.
- Keep the final PDF a sane size — a 60 MB paper full of uncompressed photos
strains the system and the reviewers' patience even where no hard cap bites.
Dual-submission and the RA-L boundary
The same manuscript may not be simultaneously under review at ICRA and anywhere
else — including RA-L, arXiv exempted (a preprint is allowed; check the current
policy text for wording). The clean patterns:
- ICRA reject in January → revise → RA-L in February. Legal and common.
- RA-L accept in autumn → transfer within the posted window for conference
presentation. Legal; this is the designed pathway.
- Same PDF to ICRA (September) and RA-L (October) "to hedge": dual submission,
discoverable trivially since both run on PaperPlaza infrastructure with
overlapping editorial boards.
Metadata fields that outlive the deadline
Three PaperPlaza fields have consequences authors rarely anticipate:
- Author list: feeds the eventual IEEE Xplore record; a co-author omitted
"to add later" creates an authorship dispute vector, not a placeholder.
- Keywords: route the paper to an Editor's area and its reviewer pool —
effectively the only reviewer-selection influence authors get (see
icra-review-process).
- Abstract: printed in the program and reviewer-bidding views; a stale
abstract that diverges from the PDF makes bidding reviewers feel misled.
Return-without-review triage
severity ▲
│ 7-page content body ............... certain return, no repair after deadline
│ author names on page 1 (2026+) .... administrative reject
│ template swapped for ML style ..... compliance flag, likely return
│ references interleaved so page 7
│ contains prose ................... counts against content limit — fix now
│ video 200 s / 25 MB ............... attachment refused; trim or re-encode
│ acknowledgements naming the grant
│ and PI .......................... anonymity leak; delete for review
▼ metadata abstract ≠ PDF abstract ... sloppiness signal to the AE; sync them
Pre-upload sequence (final 48 hours)
- Regenerate the PDF from a clean clone of the paper repo; confirm the IEEE
template class file is untouched.
- Run the page split check: content ends on or before page 6; pages 7-8 contain
nothing but the bibliography.
- Anonymity sweep: PDF metadata (
pdfinfo), acknowledgements, self-citations,
figure backgrounds, video frames, embedded links.
- Play the video attachment start to finish on a machine without the lab
codecs; verify duration, size, and that captions carry no names.
- Enter/verify PaperPlaza metadata: authors, keywords, abstract synced to PDF.
- Upload paper and video; pass the compliance check; archive the confirmation.
Output format
[ICRA upload verdict] go / fix-first / no-go
[Page audit] content <n>/6, reference pages <n>/2
[Anonymity audit] pass / leaks: <list>
[Video audit] <s>s / <MB>MB / format / window date ok?
[Metadata audit] authors final? keywords chosen? abstract synced?
[Dual-submission] clear / conflict: <venue>
[Deadline math] <local time> vs 23:59 Pacific = <hours remaining>
1---2name: icra-submission3description: Use when auditing an ICRA paper for upload — PaperPlaza account and metadata, the 6-page-plus-2-reference-page limit, IEEE two-column template compliance, the double-anonymous rule introduced in the 2026 cycle, video-attachment specs (180 s, 20 MB), dual-submission conflicts with RA-L, and return-without-review triggers.4metadata:5 source: https://skillmd.com/skills/brycewang-stanford/icra-submission6---78# ICRA Submission910Final-gate audit for a direct ICRA conference submission on PaperPlaza11(`ras.papercept.net`), the RAS submission system used across ICRA, IROS, and RA-L.12Everything here is anchored to the verified ICRA 2026 cycle; the numbers move, so13open the current year's "call for papers" page before trusting any of them.1415## Hard limits (2026-cycle values — reverify each year)1617| Check | 2026 rule | Consequence if violated |18|---|---|---|19| Page count at submission | 6 pages of content + up to 2 pages containing only references | Returned without review |20| Format | IEEE two-column conference template (RAS PaperPlaza support page templates), PDF only | Compliance failure at upload |21| Anonymity | Double-anonymous: no authors/affiliations in the PDF; full author list entered in PaperPlaza metadata | Administrative rejection |22| Video attachment | ≤ 180 seconds, ≤ 20 MB, one of mpeg/mp4/mpg | Rejected by the upload form |23| Video timing | Only inside the posted windows (2026: Aug 5–Sep 9 and Sep 17–22) | No video on record, no late path |24| Deadline | Sep 15, 2025, 23:59 **Pacific** (not AoE) | Closed portal |2526The Pacific-time detail is a live trap: authors calibrated to AoE deadlines at ML27conferences lose up to five hours.2829## The 2026 anonymity turn3031Through ICRA 2025, review was single-blind — author names printed on page one.32The 2026 cycle switched to double-anonymous. Concretely:3334- Strip the author block, affiliations, and identifying acknowledgements/grant35 numbers from the PDF; keep the full, final author list in PaperPlaza metadata36 (it cannot be extended casually later — treat the metadata list as binding).37- Rewrite platform tells: "our previously developed XYZ-Arm [12]" becomes "the38 XYZ-Arm [12]" with third-person citation.39- Blur or crop lab-identifying scenery in figures *and in the video attachment* —40 posters on the wall, safety vests with logos, institute door signs.41- Anonymize linked material: a GitHub URL with a lab organization name is an42 identity leak. Use an anonymized mirror or omit the link at submission.43- Because the policy is one cycle old, confirm it still holds for the target year44 before de-anonymizing or re-anonymizing anything.4546## PaperPlaza mechanics worth knowing in advance4748- Every author needs a PaperPlaza PIN; collecting PINs from six co-authors on49 deadline night is a known time sink — do it the week the site opens.50- The system runs its own PDF compliance test (fonts embedded, page size, page51 count); a failing PDF can be re-uploaded until the deadline, so test early.52- Keywords/session topics are chosen from a fixed RAS taxonomy and steer which53 Associate Editor sees the paper — pick the two or three closest, not the54 broadest.55- The submission confirmation email is the only receipt; archive it.5657## PDF craft the compliance checker enforces5859- Fonts must be embedded — the usual failures are figures exported from60 MATLAB/matplotlib with Type 3 or missing fonts; regenerate with embedded61 TrueType/Type 1 before deadline week.62- US Letter page size per the IEEE template; a European A4 build fails quietly.63- Vector figures where possible: reviewers zoom into plots of contact forces64 and trajectories, and rasterized 300-dpi exports turn to mush at 400%.65- Keep the final PDF a sane size — a 60 MB paper full of uncompressed photos66 strains the system and the reviewers' patience even where no hard cap bites.6768## Dual-submission and the RA-L boundary6970The same manuscript may not be simultaneously under review at ICRA and anywhere71else — including RA-L, arXiv exempted (a preprint is allowed; check the current72policy text for wording). The clean patterns:7374- ICRA reject in January → revise → RA-L in February. Legal and common.75- RA-L accept in autumn → transfer within the posted window for conference76 presentation. Legal; this is the designed pathway.77- Same PDF to ICRA (September) and RA-L (October) "to hedge": dual submission,78 discoverable trivially since both run on PaperPlaza infrastructure with79 overlapping editorial boards.8081## Metadata fields that outlive the deadline8283Three PaperPlaza fields have consequences authors rarely anticipate:8485- **Author list**: feeds the eventual IEEE Xplore record; a co-author omitted86 "to add later" creates an authorship dispute vector, not a placeholder.87- **Keywords**: route the paper to an Editor's area and its reviewer pool —88 effectively the only reviewer-selection influence authors get (see89 `icra-review-process`).90- **Abstract**: printed in the program and reviewer-bidding views; a stale91 abstract that diverges from the PDF makes bidding reviewers feel misled.9293## Return-without-review triage9495```text96severity ▲97 │ 7-page content body ............... certain return, no repair after deadline98 │ author names on page 1 (2026+) .... administrative reject99 │ template swapped for ML style ..... compliance flag, likely return100 │ references interleaved so page 7101 │ contains prose ................... counts against content limit — fix now102 │ video 200 s / 25 MB ............... attachment refused; trim or re-encode103 │ acknowledgements naming the grant104 │ and PI .......................... anonymity leak; delete for review105 ▼ metadata abstract ≠ PDF abstract ... sloppiness signal to the AE; sync them106```107108## Pre-upload sequence (final 48 hours)1091101. Regenerate the PDF from a clean clone of the paper repo; confirm the IEEE111 template class file is untouched.1122. Run the page split check: content ends on or before page 6; pages 7-8 contain113 nothing but the bibliography.1143. Anonymity sweep: PDF metadata (`pdfinfo`), acknowledgements, self-citations,115 figure backgrounds, video frames, embedded links.1164. Play the video attachment start to finish on a machine without the lab117 codecs; verify duration, size, and that captions carry no names.1185. Enter/verify PaperPlaza metadata: authors, keywords, abstract synced to PDF.1196. Upload paper and video; pass the compliance check; archive the confirmation.120121## Output format122123```text124[ICRA upload verdict] go / fix-first / no-go125[Page audit] content <n>/6, reference pages <n>/2126[Anonymity audit] pass / leaks: <list>127[Video audit] <s>s / <MB>MB / format / window date ok?128[Metadata audit] authors final? keywords chosen? abstract synced?129[Dual-submission] clear / conflict: <venue>130[Deadline math] <local time> vs 23:59 Pacific = <hours remaining>131```