"Oops, I Erred" — Collecting, Curating and Using Imperfect Data for Realistic Scenarios
A workshop at CoRL 2026
November 12, 2026 · JW Marriott Austin, Austin, Texas
Robot learning has made rapid progress on datasets of clean, successful demonstrations. But a policy trained only on curated successes only knows what to do when everything goes well ... What if it doesn't?
This workshop is about other data: execution failures and near-misses, noisy or suboptimal demonstrations, out-of-distribution events, and the unexpected things people do around robots. Every lab produces it, and almost none of it is documented, shared, or reused.
We want to work out, together, how to gather this data, how to curate it, and finally, what to do with it. Can human errors in the dataset provide important context, and how can we capture this? Can imperfect data help robots succeed in imperfect settings? Which failures actually matter, and how should they be labelled? What would shared infrastructure for collecting and publishing this data across academia and industry look like?
We invite you to join us to address these questions, any many more. Come with a paper, a dataset, or just an opinion. Alongside talks and a panel, a collaboration session and poster session give everyone room to argue about all things imperfect data directly. We will fold the results into a community white-paper and share your input with the rest of the CoRL community.
Last but not least, we are working on a format for you to directly submit and present imperfect data, and maybe get an award for the most hilarious failure.
We meet on Thursday, November 12, 2026 at the JW Marriott Austin in Austin, Texas, the workshop day following the main CoRL conference.The exact schedule is still being worked out; see the schedule for details once available, and the call for contributions to take part.
