Compare access, license, modality, signal, and scale before a training run.
Showing 42 of 42 datasets · 29 open · 13 restricted · 22 on Hugging Face · published catalog
Large-scale real-world humanoid interaction data
Humanoid VLA training
Non-commercial share-alike license; commercial use requires separate legal review or permission.
policy learning
policy learning
Verify the selected release and downstream license.
Language- and goal-conditioned manipulation
Open-vocabulary policies
A core public baseline for VLA and policy post-training.
In-the-wild real-world robot manipulation
VLA fine-tuning
Strong real-world diversity; verify release subset completeness before processing.
Synchronized first-person and third-person skilled activity
Learning from skilled human demonstrations
Especially relevant when a task needs both wearable camera context and external validation views.
Large-scale daily-life egocentric video
Pretraining egocentric perception
Best treated as a large source corpus rather than a direct robot-action dataset.
Ego4D-style activity sample with narrations
Activity-recognition examples
Good open sample for testing language annotations around egocentric clips.
Apache-2.0 LeRobot bottle-adjustment task
LeRobot conversion tests
One of the more directly robot-task-shaped open HF entries.
First-person affordance labels for manipulation
Affordance pretraining
Useful for dataset curation labels even when source videos are not robot trajectories.
Hand visibility and active-manipulation benchmark
Benchmarking egocentric data quality
Good open-access companion to the gated Egocentric-10K source dataset.
Long-form egocentric video QA
Video-language model evaluation
Not a manipulation training set, but helpful for evaluating whether agents understand long first-person context.
Catalog metadata for first-person manipulation datasets
Dataset registry examples
Best treated as a registry source rather than raw training data.
GoPro first-person pick-and-place trajectories
Pick-and-place benchmark fixtures
Open access but non-commercial license terms apply.
Long-term object tracking in egocentric video
Object persistence in robot tasks
Useful for Bench/Data integration when failures involve losing an object through a manipulation step.
Interleaved embodied reasoning and robot-control pretraining
Embodied reasoning
Separate EO-Data1.5M training samples from EO-Bench evaluation data.
VLA training
VLA training
Verify the selected release and downstream license.
Unscripted kitchen actions from wearable cameras
Kitchen manipulation vocabulary
Still a strong reference point for kitchen-centric manipulation data.
Long-horizon mobile dual-arm manipulation
Mobile manipulation
Strong source for connecting subtask annotation to mobile manipulation evaluation.
Real-robot bimanual evaluation benchmark
Real-robot VLA evaluation
Author-created benchmark; reported model results are not independent OpenBot reproduction.
CC-BY household first-person manipulation samples
Residential manipulation examples
Small open-access sample, useful for demos rather than model-scale training.
MIT-licensed industrial first-person video samples
Industrial manipulation examples
Useful for examples and smoke tests; scale is intentionally small.
MIT-licensed preview of egocentric action annotations
Action-label schema tests
Preview-scale dataset, not the full corpus.
Community robot demonstrations in a shared training format
Open VLA pretraining
Licenses and annotation quality vary by repository; treat this as a collection with per-source governance.
Long-horizon manipulation post-training and evaluation
Long-horizon policy learning
Track as a LIBERO dataset release, not as an unrelated top-level series once versioned Catalog storage is active.
Household manipulation with pose annotations
Household skill segmentation
One of the more directly robotics-relevant egocentric datasets in this catalog.
Egocentric dexterous manipulation sample set
Dexterous-hand data demos
Open access on Hugging Face, but non-commercial license terms apply.
Cross-embodiment robot learning in a unified format
Cross-embodiment VLA pretraining
Licenses differ across constituent datasets and must be checked individually.
MIT-licensed POV robotics video samples
Small open fixture for first-person video ingestion
Small sample dataset, not a large training corpus.
Multi-embodiment demonstrations with explicit failures
Failure mining
One of the strongest current matches for OpenBot loop-signal analysis.
Bimanual simulation data generation and evaluation
Bimanual VLA post-training
This entry covers a generator, released trajectories, and benchmark; keep those artifacts distinct in the future release schema.
Egocentric video QA for wearable assistants
Agent evaluation over streaming first-person video
Less robot-action focused, but useful for evaluating an agent's temporal grounding.
Multimodal human experience for embodied AI
World model pretraining
Very large and controlled-access; the practical OpenBot path is metadata indexing plus targeted subset pulls.
Multiview egocentric scene reconstruction
View synthesis for wrist/ego cameras
Short recordings, but high camera density and strong reconstruction value.
Small egocentric sample set for manipulation domains
Fast catalog smoke tests
Small but concrete; useful for demos and ingestion tests.
Cross-dataset active labor and hand-presence evaluation
Comparing egocentric corpora before ingest
Open evaluation layer; the larger source datasets may still be gated.
Factory worker first-person video
Industrial manipulation priors
Good candidate for OpenBot Data indexing, dedup, and active-manipulation scoring.
Bimanual manipulation in LeRobot format
Schema validation for LeRobot v3
Small, but excellent for testing catalog pages, loaders, and schema conversions.
Metric RGB-D pretraining for spatial perception
Depth completion
Perception dataset rather than an action-policy training corpus; RobbyVla is the manipulation-related subset.
LeRobot egocentric hand-tracking sample
Testing LeRobot import paths
Good open sample for OpenBot Data loaders and dataset pages.
Contact-rich multimodal manipulation paired with human demonstrations
Contact-rich policy learning
Commercial-use rights differ by subset; privacy-sensitive human recordings are included.
Video + audio + IMU egocentric clips
Wearable-assistant evaluation
A practical bridge between video-only egocentric data and robot-ready sensor streams.
VLA training
VLA training
Verify the selected release and downstream license.
OpenBot Data turns raw sources into versioned, replay-ready training data.
