MolmoAct2-BimanualYAM: Bimanual Robot Manipulation Demonstrations

MolmoAct2-BimanualYAM is a large-scale collection of bimanual robot manipulation demonstrations created using LeRobot. The dataset contains more than 720 hours of training demonstrations for diverse tabletop manipulation tasks. It was created by AllenAI and was last updated on the platform in June 2026.

Use Cases

• Training imitation learning models based on bimanual robot demonstrations.
• Developing robot policies for tabletop manipulation tasks based on the described task diversity.
• Benchmarking robot learning algorithms based on the scale of demonstration data.
• Researching language-conditioned robot control based on the mentioned language annotations.

Strengths

• Contains more than 720 hours of robot demonstration data.
• Focuses on bimanual manipulation, a specific and complex robotic domain.
• Covers diverse tabletop manipulation tasks.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and file formats are unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

AllenAI via Hugging Face.

Collection Method

Created using LeRobot; likely contains robot demonstration recordings.

Freshness

Last updated 2026-06-09 23:47:27; freshness should be verified.