720 hours of bimanual robot manipulation demonstrations for tabletop tasks, collected for the MolmoAct2 project. This subset includes annotated language instructions. The dataset was created by AllenAI using LeRobot and was last updated on May 6, —.
Use Cases
- Training imitation learning models based on the large-scale collection of robot demonstrations.
- Developing language-conditioned robot policies based on the annotated language instructions.
- Benchmarking bimanual manipulation algorithms based on the diverse tabletop tasks.
- Researching robot skill generalization based on the extensive demonstration hours.
Strengths
- Contains more than 720 hours of training demonstrations, providing substantial scale.
- Focuses on bimanual manipulation, a specific and complex robotic domain.
- Includes language annotations, adding a semantic layer to the demonstrations.
Limitations
- Description metadata is limited; actual data quality requires manual inspection after download.
- Column-level documentation is absent; field semantics must be inferred after download.
- Row count, file formats, and license are unknown, which may limit suitability assessment.
Provenance
- Source
- AllenAI
- Collection Method
- Created using LeRobot, likely from recorded robot demonstrations.
- Time Range
- null
- Freshness
- Last updated 2026-05-06 06:35:12; freshness should be verified.
- Geography
- null