Evorl Plug Stack: Robot Teleoperation Demonstrations for Reinforcement Learning

A dataset of robot teleoperation demonstrations created using the LeRobot framework. It contains 20 episodes totaling 7,487 frames of data for a single task, likely involving a 'so_follower' robot type. The dataset was uploaded by author SL7iii to Hugging Face and was last updated on July 6, 2026.

Use Cases

• Training imitation learning policies based on recorded human teleoperation demonstrations.
• Benchmarking reinforcement learning algorithms on real-world robot manipulation tasks.
• Developing visuomotor control models based on the multimodal data likely present in the episodes.

Strengths

• Contains 7,487 frames of demonstration data across 20 episodes, providing a substantial base for learning.
• Data is structured into chunks of 1,000 frames, which may facilitate efficient data loading and processing.
• Includes video data recorded at 30 frames per second, offering a high-fidelity visual record of robot actions.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• The dataset description is minimal; actual data quality and content require manual inspection after download.

Provenance

Source

huggingface

Collection Method

Created using the LeRobot framework, likely involving human teleoperation.

Freshness

Last updated 2026-07-06 14:48:07; freshness should be verified.