Eval Pi0Test0 Ae 16B 10Epch 0: Robot Teleoperation Episodes for Imitation Learning

A dataset of 5 robot teleoperation episodes, comprising 8,398 video frames, created using the LeRobot framework. It was authored by asdl-unist and last updated on June 20, 2026. The data is structured for machine learning, with splits defined for training.

Use Cases

• Training robot control policies based on teleoperated demonstration data mentioned in the description
• Evaluating imitation learning algorithms on a fixed set of 5 episodes
• Benchmarking model performance on the 'so_follower' robot type specified in the metadata

Strengths

• Contains 8,398 video frames across 5 episodes, providing a base for sequential learning tasks
• Metadata specifies a structured format with defined splits and a robot type, aiding reproducibility
• Video data is recorded at 30 frames per second, which is a standard rate for motion capture

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 is unknown, which may limit suitability assessment

Provenance

Source

huggingface

Collection Method

Created using the LeRobot framework, likely involving human teleoperation.

Freshness

Last updated 2026-06-20 14:19:17; freshness should be verified