Hugging Face's logo

Datasets:
zjt24
/
aloha_static_fork_pick_up

Tasks:
Robotics
Tags:
LeRobot
robotics
time-series
tsfile
License:
Dataset card Files
xet
 Community
aloha_static_fork_pick_up / README.md
zjt24's picture
zjt24
Add TsFile (converted from lerobot/aloha_static_fork_pick_up)
bd168c5 verified
preview code
|
Raw
History Blame Contribute Delete
3.34 kB
metadata 
license: apache-2.0
task_categories:
  - robotics
tags:
  - LeRobot
  - robotics
  - time-series
  - tsfile
modality: timeseries
pretty_name: aloha_static_fork_pick_up TsFile
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/aloha_static_fork_pick_up.tsfile

aloha_static_fork_pick_up TsFile

This dataset is an Apache TsFile conversion of the Hugging Face dataset lerobot/aloha_static_fork_pick_up. The source dataset was created with LeRobot and is licensed under Apache 2.0.

Modalities: Time-series. The original visual MP4 streams remain available in the source dataset; this repository stores the numeric robot state, effort, action, frame metadata, task index, and episode tags as TsFile.

Source Dataset

  • Original dataset: lerobot/aloha_static_fork_pick_up
  • Source task: "Pick up the fork and place it on the plate."
  • Codebase version: LeRobot v3.0
  • Robot type: unknown
  • Split: train (0:100)
  • Episodes: 100
  • Frames: 60,000
  • Sampling rate: 50 fps
  • Tasks: 1
  • Source data path: data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet
  • Source video path: videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4

Converted Files

  • TsFile: data/aloha_static_fork_pick_up.tsfile
  • Rows: 60,000
  • Table name: aloha_static_fork_pick_up
  • Time precision: milliseconds
  • Mirrored metadata: meta/, with meta/info.json updated for the TsFile artifact

Schema

Time is generated as round(timestamp * 1000) milliseconds. Time restarts within each episode, and devices are identified by the original LeRobot tag columns.

  • TAG columns: episode_index, task_index
  • FIELD metadata columns: frame_index, sample_index, next_done
  • FIELD vectors: observation_state_0..observation_state_13, observation_effort_0..observation_effort_13, and action_0..action_13

The next_done field is stored as an INT64 flag: 0 means false and 1 means true. The converted data has 100 true flags, one terminal frame per episode.

Conversion Notes

  • The source timestamp column is dropped after being mapped to Time; it is recoverable as Time / 1000 seconds.
  • The source index column is renamed to sample_index.
  • The source next.done column is renamed to next_done and stored as 0/1 for stable TsFile readback.
  • Vector columns are flattened by preserving the full source column name, replacing . with _, and appending the element index.
  • Source video features are not uploaded here: observation.images.cam_high, observation.images.cam_left_wrist, observation.images.cam_low, and observation.images.cam_right_wrist. Use the original dataset's videos/ tree for the visual streams.
  • Aside from the redundant timestamp column and omitted video files, no numeric time-series rows are intentionally dropped.

Read Example 

from tsfile import TsFileReader

path = "data/aloha_static_fork_pick_up.tsfile"

with TsFileReader(path) as reader:
    tables = reader.get_all_table_schemas()
    print(tables.keys())