Call for Papers

Sim2Real2Field

Sim-to-Real-to-Field: The Science of Transfer

CoRL 2026 Workshop · 2026 (date TBA) · location TBA

From simulation, to the lab, to robust deployment in the field.

Submit on OpenReview →

About

Robot learning has made remarkable progress at crossing the reality gap: we train in simulation, where data is cheap, exploration is safe, and scale is attainable, then transfer to physical hardware. But “real” usually means a single robot in a controlled lab, and the transfer is not finished there. A lab demonstration certifies a point; deployment in the field demands robustness measured distributionally — across many robots, sites, and operating conditions, where rare failure modes rather than average performance decide success.

We call this fuller arc sim-to-real-to-field (SRF). Robot learning is our center of gravity; we draw selectively on adjacent sim-to-real fields — scientific machine learning, control, and systems — for methods and evaluation standards the robotics community can borrow and lend.

Three driving questions

  1. Q1 — Training environments to field deployment. How do we design and combine simulation environments — across fidelity, diversity, and abstraction — so that what robots learn in simulation remains useful in the field?
  2. Q2 — Lab metric to field guarantee. How do we scope, monitor, and survive transfer — characterizing the operating distribution, estimating achieved transfer from onboard signals alone, and staying safe under transfer failure?
  3. Q3 — Adaptation algorithms for field transfer. What lets a deployed policy adapt online to the distribution shifts it meets in the wild — and can foundation models help handle shifts unanticipated before deployment?

Call for Papers

We host an open call for contributed papers, inviting submissions that foster discussion and exploration of emerging ideas in sim-to-real transfer. We particularly encourage early-stage research, novel problem formulations, negative results, lessons learned from practical deployments, and system-building experiences. Selected submissions are featured as interactive poster sessions to maximize discussion.

Before you submit

  • All authors must have an OpenReview profile. Per OpenReview: new profiles created without an institutional email go through a moderation process that can take up to two weeks — create your profile well ahead of the deadline.
  • Reviewing commitment. At least one author of each submitted paper must be willing to review for the workshop, with a workload of at most 2 papers.

Submit on OpenReview →

Submissions are short and non-archival. We discourage papers already accepted to the main CoRL 2026 conference. Reviewing follows standard CoRL guidelines. Deadlines will be posted here.

Program

A half-day, in-person workshop built around exchange — less than 25% of the program is talks. Schedule is tentative and subject to change.

0:00Opening: framing the three questions + polling
0:10Invited talks — Amy Zhang (UT Austin) & Zhongyu Li (CUHK)
0:55Invited talks — Claire Tomlin (UC Berkeley) & Michael Lutter (Boston Dynamics)
1:25Poster session & coffee; cross-domain matchmaking; mentoring corner
2:05Breakouts on Q1–Q3 (mixed-domain) → report-back
2:50Fishbowl panel: what predicts transfer — scale, metrics, or structure?
3:35Closing & community next steps

Speakers

Four invited speakers spanning career stages, academia and industry, and regions. Each addresses elements of Q1–Q3 and how the field should move forward.

Organizers

Information

VenueCoRL 2026 (location TBA)
Date2026 (TBA)
FormatHalf-day, in-person
ProceedingsNon-archival

Contact: rosario@cs.washington.edu