brycewang-stanford IROS Author Response
Use this after IROS reviews arrive — and, crucially, before submission too, because IROS traditionally offers no rebuttal. The reviews come with the decision, so the "response" is spread across three real channels: pre-answering in the paper, editing the camera-ready, and, if rejected or on the journal route, a response letter or resubmission memo. Confirm the current cycle's policy before assuming a rebuttal window does or does not exist.
Channel 1: pre-answer before you submit
The most effective IROS author response is written before anyone reviews the paper. Anticipate the standard objections and disarm them in the body and the video:
- "Did it really work?" → shown, uncut trials plus a labeled failure in the video.
- "Is it reliable or cherry-picked?" → trial counts, resets, and a failure taxonomy in the body.
- "Does sim imply real?" → a stated sim-to-real gap, not an implied zero.
- "Is the baseline fair?" → same-platform baseline with comparable tuning.
Channel 2: the camera-ready revision
When the decision is accept (often with requested changes), the meta-review is your instruction set. Integrate its asks without inflating the contribution beyond what was reviewed.
| Meta-review ask | Camera-ready move | Trap to avoid |
|---|---|---|
| "Report failure cases" | Add the failure taxonomy table you logged | Do not add new claims the reviewers never saw |
| "Clarify the compute budget" | State measured rate and power on the robot | Do not quietly change the reported system |
| "Compare to system X" | Add the same-platform comparison if you have it | Do not fabricate a comparison you did not run |
| "Fix anonymized links" | Replace with the public, licensed release | Do not leave "code coming soon" |
Channel 3: RA-L response letters and resubmission memos
- If your work is on the RA-L journal pathway, that review does have response letters: answer each point, quote the change, and cite the revised line — a very different discipline from the no-rebuttal conference.
- If the IROS paper is rejected, write a short internal resubmission memo: which objections were evidence gaps (fixable with a run), which were framing, and which venue (next ICRA, CoRL, or RA-L) the revised paper best fits.
Micro-example: turning a rejection into a plan
Reviewers rejected a manipulation paper for "insufficient real-robot evidence." The memo:
- Name the gap precisely — 20 trials on 3 objects is too thin for the reliability claim.
- Scope the fix — 120 trials across 15 objects with a logged failure taxonomy.
- Choose the target — RA-L (for revision cycles) or the next ICRA (for the spring deadline).
- Decide the video re-cut — add a labeled failure to pre-answer the same objection.
Response draft skeleton (camera-ready or RA-L letter):
[Point] <reviewer/meta-review concern>
[Change] <exact edit or added table/figure>
[Location] <section / table / line in the revised paper>
[Not added] <new claims deliberately withheld to stay within review scope>
Output format
[Channel] pre-submission / camera-ready / RA-L letter / resubmission memo
[Priority concern] <the objection that most threatens acceptance>
[Response] <IROS-appropriate, scoped, anonymity-safe>
[Evidence anchor] <paper/video/log item>
[Forbidden] <new unsupported claims / identity leaks / fabricated comparison>