DATARESEARCH PREVIEW
DROID collaboration
Everyday manipulation. Real environments.
Stereo RGBRobot actionsLanguage
76k trajectories564 scenes
DATA FOR THE PHYSICAL WORLD
The next move
starts with
the right data.
Real-world footage. Robot demonstrations.
Find your next dataset, or put yours on the map.
Open research. Real demonstrations.
Start with a sample.
Start with a sample.
FROM THE REAL WORLD01 / 04
HUMAN DEMONSTRATIONSEvery action is a
Every action is a
starting point.
UMI · Public research demoView source
EXPLORE OPEN RESEARCH FROMDROIDCOLLABORATIONBerkeleyRAIL LABUMIRESEARCH PROJECTOpen XEMBODIMENTIndependent examples.
No affiliation implied.
No affiliation implied.
THE CATALOG
Less searching. More finding.
Explore public datasets and see what the data actually looks like.
⌘ K
6 collections
DATARESEARCH PREVIEW
UC Berkeley · RAIL
A broad collection of everyday robot skills.
RGB videoRobot actionsLanguage
60,096 trajectories24 environments
DATAFIRST PERSON
Stanford · Columbia · TRI
Human demonstrations, from a new perspective.
Wrist cameraGripper motionHuman demos
In the wild collectionHandheld grippers
DATARESEARCH PREVIEW
Open X-Embodiment collaboration
Experience shared across different robots.
Multiple robotsRobot actionsRLDS
1M+ episodes22 embodiments
DATARESEARCH PREVIEW
From BridgeData V2
A closer look at deformable-object tasks.
Deformable objectsRGB videoManipulation
Task preview cloth foldingFrom BridgeData V2
DATAFIRST PERSON
From the UMI project
Two hands. One coordinated task.
BimanualHuman demosWrist camera
Task preview bimanual collectionFrom the UMI project
4 original sources · 6 demo collectionsFree access through the original providers. Source terms apply.
ON THE OTHER SIDE OF THE CAMERA?
You have the data.
Let’s get it out there.
Start with a sample. Tell us what you have, how much is available, and when it’s ready.
A sample is all it takes to get started. YOUR NEXT OPPORTUNITY
Something worth sharing.
A SHORT PATH FROM SAMPLE TO DATA
Keep it moving.
01
Find your starting point
Browse by task, capture setup, or environment. Save the datasets that catch your eye.
02
See the actual sample
Watch the footage and check the details. Know what’s in the dataset before going further.
03
Get the data
Access open datasets at the source, or tell us what you’re looking for.