Iros workflow

IROS Workflow

Use this as the project-management skill for an IROS submission. Replace every date with the current official timetable and work backward from PaperPlaza cutoffs.

IROS is a conference, not a journal: it has no standing editor-in-chief and no article-processing charge. It is co-sponsored by IEEE RAS and the Robotics Society of Japan (RSJ) — hence "IEEE/RSJ" — and the organizing committee rotates yearly between regions, so re-check the current CFP and committee page rather than carrying names or rules forward. The cost model is registration plus any per-page charge, and proceedings go to IEEE Xplore.

The fall-venue rhythm (2026 shape)

IROS is the fall IEEE-RAS mega-venue, the seasonal counterpart to ICRA's spring cycle. The 2026 edition ran a spring submission into an autumn conference:

• Paper deadline in early spring (2026: March 2).
• Video attachment a few days later (2026: March 5) — a distinct cutoff, not tied to the paper.
• A long summer review silence, then notification in mid-June (2026: June 16).
• Camera-ready in July (2026: July 10), then the conference in late September/October (2026: Sept 27 - Oct 1, Pittsburgh).

Knowing you are on the fall calendar changes routing: a project that misses IROS's spring deadline has CoRL and the next ICRA as nearer alternatives than waiting a full IROS year.

Milestones

• Venue fit: confirm the contribution is an integrated intelligent system with credible robot evidence.
• Evidence lock: freeze the platform, task distribution, trial protocol, baselines, and the video plan.
• Paper deadline: upload the anonymous PDF (references counted in the page budget) and PaperPlaza fields.
• Video deadline: upload the ≤ 60 s / ≤ 10 MB attachment by its own cutoff.
• Review silence: nothing to do but keep the robot and logs intact for camera-ready.
• Notification: triage reviews; there is traditionally no rebuttal, so plan camera-ready edits instead.
• Camera-ready: de-anonymize, integrate meta-review asks, file the IEEE copyright form, register.
• Presentation: at least one author presents in person; prepare the talk and any live/updated video.

Backward plan from the paper deadline

Weeks out (heuristic)	Real-robot systems milestone
10+	Platform frozen; task distribution and success criterion defined
8	Baselines implemented on the same hardware for a fair comparison
6	Main trials run, resets logged, failures categorized
4	Sim-to-real study complete; the gap is a number, not a hope
3	Full draft in IEEE two-column format; references budgeted as content
2	Video footage banked from real sessions; rough 60 s cut assembled
1	Anonymity sweep across PDF, figures, and video; PaperPlaza PINs collected
0	Paper uploaded; video by its separate cutoff

These offsets are planning heuristics only — anchor each to the current official timetable, never to a previous cycle's calendar.

Failure modes by stage

• Filming on deadline week produces worse footage and worse anonymity than banking clips all along.
• Treating references as free space overruns the budget, because IROS counts them in.
• Assuming a rebuttal exists wastes the notification window; IROS traditionally ships reviews with the decision, so the only lever is camera-ready.
• Letting the robot get reconfigured before camera-ready makes a requested extra trial impossible.

Coordination notes

• Assign one named owner for the video and one for the anonymity sweep; shared ownership is how both slip.
• Archive the exact submitted PDF and video, since the camera-ready must trace to the reviewed version.

Output format

[Current stage] idea / experiments / writing / submission / video / review / notified / camera-ready
[Next official deadline] <date and source, or unknown>
[Critical path] <three tasks that determine readiness>
[Risk register] <page-budget/anonymity/robot-evidence/video/transfer>
[Owner map] <task -> person or role>