Push Panda Toilet Paper S2 Postprocessed: Robot Manipulation Videos

NONHUMAN-RESEARCH provides the post-processed LeRobot dataset for the task of pushing a panda toy and toilet paper. The dataset contains 25 episodes and is structured under the LeRobot v2.1 format with standard meta/, data/, and videos/ directories. A lightweight preview subset is configured as the default view on the Hugging Face Dataset Viewer.

Use Cases

• Train robot control policies based on demonstration videos.
• Benchmark robot manipulation performance on the specific task of pushing articulated objects.
• Develop computer vision models for tracking objects in robot interaction videos.
• Study imitation learning techniques using the structured episode data.

Strengths

• Dataset is structured in the standard LeRobot v2.1 format, which suggests a consistent organization.
• Includes a lightweight preview subset configured for immediate visualization on the dataset platform.

Limitations

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

Provenance

Source

NONHUMAN-RESEARCH via Hugging Face.

Collection Method

Post-processed from the LeRobot dataset.

Freshness

Last updated 2026-05-15 17:59:48; freshness should be verified.