README.md

metadata

license: apache-2.0
task_categories:
  - robotics
tags:
  - LeRobot
  - TsFile
  - time-series
  - robotics
modality:
  - timeseries
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/eval1_fixed_all_v2.tsfile

eval1_fixed_all_v2

This dataset is converted from robot-learning-group47/eval1_fixed_all_v2 to Apache TsFile format.

Modalities: Time-series. The original dataset also contains a front-camera video stream, which remains in the source Hugging Face dataset and is not included in this converted repository.

Source Dataset

• Original dataset: robot-learning-group47/eval1_fixed_all_v2
• Format: LeRobot v3.0
• Robot type: so_follower
• License: apache-2.0
• Scale: 61 episodes, 10,980 frames, 3 tasks, 15 fps
• Original data path: data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet
• Original video path: videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4

Converted Files

• TsFile data: data/eval1_fixed_all_v2.tsfile
• Mirrored metadata: meta/
• Rows in TsFile: 10,980
• Time precision: milliseconds

Schema

Time is synthesized as round(timestamp * 1000) and restarts per episode.

TAG columns:

• episode_index
• task_index

FIELD columns:

• frame_index
• sample_index renamed from source index
• action_0 through action_5, flattened from action[6]
• observation_state_0 through observation_state_5, flattened from observation.state[6]

Flattened vector values are stored as single-precision FLOAT fields.

Conversion Notes

• The source timestamp column is not retained as a separate field because it is redundant with Time / 1000 seconds.
• The source front-camera video stream observation.images.front is not uploaded here. Use the original dataset for videos: https://huggingface.co/datasets/robot-learning-group47/eval1_fixed_all_v2/tree/main/videos
• meta/info.json is mirrored from the source and updated to describe the converted TsFile schema, original video path, row count, TAG columns, flattened features, and video policy.
• Aside from the redundant timestamp field and omitted video stream, no rows or numeric time-series fields are intentionally dropped.

Read Example

# Read data/eval1_fixed_all_v2.tsfile with the Apache TsFile Java/Python SDK.
# Query one episode with a predicate such as:
# WHERE episode_index = 0