iamlab_cmu_pickup_insert_train_500_631_augmented (TsFile)

This repository contains a TsFile conversion of the OXE-Aug / LeRobot dataset oxe-auge/iamlab_cmu_pickup_insert_train_500_631_augmented. It keeps the numeric robot time-series data and LeRobot metadata, while the large video assets remain in the original Hugging Face dataset.

Source Dataset

• Original dataset: oxe-auge/iamlab_cmu_pickup_insert_train_500_631_augmented
• Format: OXE-Aug / LeRobot v2.1
• License: cc-by-4.0
• Task category: robotics
• Modalities in this conversion: Time-series
• OXE-Aug website: https://oxe-aug.github.io/
• Project repository: https://github.com/GuanhuaJi/oxe-aug
• Paper: https://openreview.net/forum?id=WuBv9-IGDUA

Dataset Scale

Item	Value
Episodes	131
Frames / TsFile rows	30,143
Tasks	7
Sampling rate	20 fps
Source video files	1,179
Uploaded TsFile files	1
Converted columns	246 including Time

Task labels from the source metadata:

task_index	task
0	Pick up blue block.
1	Insert greeen block.
2	Pick up yellow block.
3	Pick up green block.
4	Pick up pink block.
5	Pick up pink flower.
6	Pick up narrow blue stick.

Data Files

The converted time-series data is stored as:

data/iamlab_cmu_pickup_insert_train_500_631_augmented.tsfile

The source LeRobot layout was:

data_path : data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet
video_path: videos/chunk-{episode_chunk:03d}/{video_key}/episode_{episode_index:06d}.mp4

The repository also includes converted LeRobot metadata under meta/. The meta/info.json file was rewritten so that data_path points to the TsFile artifact and tsfile_conversion documents the conversion.

TsFile Schema

The TsFile table name is:

iamlab_cmu_pickup_insert_train_500_631_augmented

Column roles:

• Time: TsFile timestamp in milliseconds. It is computed as round(timestamp * 1000) and restarts per episode.
• TAG columns: episode_index, task_index. These identify the TsFile devices / task groups and allow queries such as filtering a single episode or task.
• FIELD columns: all remaining scalar numeric/string measurements, including frame_index, sample_index, natural_language_instruction, robot joint states, end-effector poses, base positions, base orientations, and error signals.

The original index source column is renamed to sample_index.

Conversion Notes

This conversion used the generic LeRobot converter in the huggingface_to_tsfile pipeline.

• Vector/list columns are flattened into scalar FLOAT fields. The full source column name is preserved, . is replaced with _, and an element index is appended. For example, observation.state becomes observation_state_0 through observation_state_19.
• The source timestamp column is not kept as a separate FIELD because it is exactly represented by Time / 1000 seconds after conversion.
• The source index column is renamed to sample_index.
• Source video/image features are not converted into TsFile and are not uploaded here. They remain available in the original dataset under videos/.
• All episodes are merged into one TsFile. episode_index and task_index are retained as TAG columns to preserve episode/task identity.

Dropped or omitted source features:

• Dropped as redundant time source: timestamp
• Omitted video features: observation.images.widowX, observation.images.sawyer, observation.images.ur5e, observation.images.google_robot, observation.images.jaco, observation.images.kinova3, observation.images.kuka_iiwa, observation.images.xarm7, observation.images.image

Minimal Read Example

Use an Apache TsFile reader to query:

data/iamlab_cmu_pickup_insert_train_500_631_augmented.tsfile

Typical query pattern:

SELECT *
FROM iamlab_cmu_pickup_insert_train_500_631_augmented
WHERE episode_index = 0

Citation Policy

If you use this converted dataset, cite the original upstream dataset and OXE-Aug.

Upstream Dataset Citation

@inproceedings{saxena2023multiresolution,
    title={Multi-Resolution Sensing for Real-Time Control with Vision-Language Models},
    author={Saumya Saxena and Mohit Sharma and Oliver Kroemer},
    booktitle={7th Annual Conference on Robot Learning},
    year={2023},
    url={https://openreview.net/forum?id=WuBv9-IGDUA}
}

OXE-Aug Citation

@misc{
  ji2025oxeaug,
  title  = {OXE-Aug: A Large-Scale Robot Augmentation of OXE for Scaling Cross-Embodiment Policy Learning},
  author = {Ji, Guanhua and Polavaram, Harsha and Chen, Lawrence Yunliang and Bajamahal, Sandeep and Ma, Zehan and Adebola, Simeon and Xu, Chenfeng and Goldberg, Ken},
  year   = {2025},
  note   = {Manuscript}
}