OpenBot
Back to datasets
Manipulation datasetOpenReadiness 73 · confidence 65

AgiBotWorld2026 Dataset

policy learning

12.78TBrepository size

AgiBot World 2026 Real-World Embodied Intelligence Dataset Overview As robotics research advances into real-world scenarios, the demand for authentic, high-quality data has become increasingly urgent. Following AGIBOT WORLD's "ImageNet moment," we now release the AGIBOT WORLD 2026 dataset. Built upon massive real-world scenes, it systematically spans pivotal research directions in embodied intelligence, designed to power the next generation of… See the full description on the dataset page: https://huggingface.co/datasets/agibot-world/AgiBotWorld2026.

Best for

policy learning

Not for / blocker

No verified blocker; unresolved facts remain unknown.

Download decision

Start with an official sample or one task; do not download the full release before schema validation.

Policy learningUseful

Has observation, action/state proxy, and task or language context.

fit 85 · confidence 55

World modelUseful

Has visual observations plus geometry, calibration, depth, or reconstruction cues.

fit 82 · confidence 55

WAMUseful

Contains observation, intent, action/state, and feedback-like supervision.

fit 88 · confidence 55

Verified facts and provenance

Claims, metadata verification, and sample verification are shown separately.

curated official source
Official claim · signals
VideoActionsRobot stateLanguageTask phaseSuccess FailureIntervention TakeoverRecoverySIM RealLanguage:en
Metadata verified · schema / annotations
is_result_succeedError FrameSuccess FrameIntervention Frametake_overerror_causerestorableinstruction_segmentsTask Frame2D Bounding Box
Sample / pipeline verification

Unknown — metadata conclusions do not prove sample coverage, alignment, or file integrity.

Declared loop signal coverage

Signals inferred from official metadata; Data pipeline verification is still pending.

6/7 categories present or partial

Observation / ego video

video · AgiBot World 2026 Real-World Embodied Intelligence Dataset Overview As robotics research advances into real-world scenarios, the demand for authentic, high-quality data has become increasingly urgent. Following AGIBOT WORLD's "ImageNet moment," we now release the AGIBOT WORLD 2026 dataset. Built upon massive real-world scenes, it systematically spans pivotal research directions in embodied intelligence, designed to power the next generation of… See the full description on the dataset page: https://huggingface.co/datasets/agibot-world/AgiBotWorld2026.

present

Action / hand pose / robot state

actions · robot state

present

Gaze / attention

No decision-grade evidence captured yet.

unknown

Language intent / task phase

language · task phase · language:en · task_categories:robotics

present

Feedback / correction / failure

success/failure · failure recovery

present

Sim-real pairing

sim-real · AgiBot World 2026 Real-World Embodied Intelligence Dataset Overview As robotics research advances into real-world scenarios, the demand for authentic, high-quality data has become increasingly urgent. Following AGIBOT WORLD's "ImageNet moment," we now release the AGIBOT WORLD 2026 dataset. Built upon massive real-world scenes, it systematically spans pivotal research directions in embodied intelligence, designed to power the next generation of… See the full description on the dataset page: https://huggingface.co/datasets/agibot-world/AgiBotWorld2026.

present

License / format / access

Open · cc-by-nc-sa-4.0

present

Model and task fit · OpenBot inference

Policy learningUseful

Has observation, action/state proxy, and task or language context.

fit 85 · confidence 55

World modelUseful

Has visual observations plus geometry, calibration, depth, or reconstruction cues.

fit 82 · confidence 55

WAMUseful

Contains observation, intent, action/state, and feedback-like supervision.

fit 88 · confidence 55

Failure miningUseful

Has failure/evaluation-style labels with action or manipulation context.

fit 82 · confidence 55

Good tasks

video-language reasoninglong-horizon planningskill-level imitation learningfailure and recovery mining

Blockers and unresolved evidence

  • Gaze / attentionunknown
    Not enough evidence to classify this signal. Verify metadata or a bounded sample.

Raw dataset signals

VideoActionsRobot stateLanguageTask phaseSuccess FailureIntervention TakeoverRecoverySIM RealLanguage:en

OpenBot fit

  • policy learning
  • long-horizon planning
  • skill-level imitation
  • failure recovery
  • task_categories:robotics
  • lerobot

Integration notes

    Related by signals