Paper
Submissions presenting research on imperfect data collection, curation, or usage in robot learning.
Machine Learning/CORE Unranked
Oopsie-Workshop CoRL 2026
"Oops, I Erred" — Collecting, Curating and Using Imperfect Data for Realistic Scenarios
The Oopsie-Workshop at CoRL 2026 focuses on collecting, curating, and using imperfect data—such as execution failures, noisy demonstrations, and out-of-distribution events—to train robots for realistic, imperfect real-world scenarios. The workshop aims to foster community collaboration on documenting and sharing underutilized failure data, with the goal of improving robot learning beyond curated successes.
Paper fit
Contribution paths
A strong submission should clearly identify its contribution and evaluate it appropriately.
Dataset
Submissions of imperfect data for sharing and reuse by the community.
Opinion
Contributions sharing perspectives, ideas, or proposals without formal papers or datasets.
Research areas in scope
01
Imperfect Data in Robot Learning
Execution failures and near-missesNoisy or suboptimal demonstrationsOut-of-distribution eventsUnexpected human behaviors around robotsLabeling and contextualizing failuresShared infrastructure for collecting and publishing imperfect dataCapturing human error as useful context
Policies worth checking twice
- Submissions can include papers, datasets, or opinions
- Participants are encouraged to submit imperfect data for potential presentation and awards
- Workshop outputs will be folded into a community white-paper
- Input will be shared with the broader CoRL community
Official sources
Compiled from the official call for papers. The organizers’ pages remain authoritative.
Last verified September 9, 2026