Datasets:

zjt24
/

aloha_dining_table

license: mit
task_categories:
  - robotics
language:
  - en
tags:
  - tsfile
  - time-series
  - robotics
  - lerobot
  - aloha
  - manipulation
modality: timeseries
pretty_name: ALOHA Dining Table TsFile
size_categories:
  - 10K<n<100K
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/aloha_dining_table.tsfile

ALOHA Dining Table TsFile

This dataset is a TsFile conversion of the Hugging Face dataset htrbao/aloha-dining-table, which uses the LeRobot v2.1 format.

Modalities: Time-series. The original dataset also includes three video streams, observation.images.color.high, observation.images.color.wrist_right, and observation.images.color.wrist_left; those videos are not included in this repository and remain available in the source dataset.

Source Dataset

Original dataset: htrbao/aloha-dining-table
License: MIT
Format: LeRobot v2.1
Robot type: ALOHA
Task: Spread the dining mat on the table, place the plate at the center of the mat, put the spoon to the left of the plate, and place the chopsticks to the right of the plate.
Split: single train split (0:99)
Scale: 99 episodes, 80,041 frames, 1 task, 1 data chunk
Sampling rate: 15 fps
Original video features:
observation.images.color.high at 480x640x3
observation.images.color.wrist_right at 480x640x3
observation.images.color.wrist_left at 480x640x3

Converted Files

TsFile: data/aloha_dining_table.tsfile
Rows: 80,041
Episodes: 99
Tasks: 1 (task_index = 0)
Table name: aloha_dining_table
Time precision: milliseconds
Metadata: meta/ is mirrored from the source dataset, with meta/info.json rewritten to describe the TsFile artifact and conversion mapping.

Schema

Column	Role	Type	Notes
`Time`	TIME	INT64	`round(timestamp * 1000)`, in milliseconds; restarts per episode
`episode_index`	TAG	INT64	Source episode identifier
`task_index`	TAG	INT64	Source task identifier
`frame_index`	FIELD	INT64	Source frame index, preserved
`sample_index`	FIELD	INT64	Renamed from source `index`
`action_0` ... `action_13`	FIELD	FLOAT	Flattened from `action[14]`
`observation_state_0` ... `observation_state_13`	FIELD	FLOAT	Flattened from `observation.state[14]`
`observation_velocity_0` ... `observation_velocity_13`	FIELD	FLOAT	Flattened from `observation.velocity[14]`
`observation_effort_0` ... `observation_effort_13`	FIELD	FLOAT	Flattened from `observation.effort[14]`

The 14 action and observation state elements correspond to the ALOHA joint displacements named in the source metadata: left_waist, left_shoulder, left_elbow, left_forearm_roll, left_wrist_angle, left_wrist_rotate, left_gripper, right_waist, right_shoulder, right_elbow, right_forearm_roll, right_wrist_angle, right_wrist_rotate, and right_gripper.

episode_index and task_index are TAG columns, so they form the TsFile device dimension. To read one episode, filter by episode_index.

Conversion Notes

The LeRobot frame Parquet files under data/chunk-000/ were merged into one TsFile for the train split.
Vector columns were flattened by preserving the source column name, replacing . with _, and appending the element index.
The source timestamp column is dropped because it is redundant with Time / 1000 seconds.
The source index column is renamed to sample_index.
Videos are not uploaded here. Use the original dataset videos: https://huggingface.co/datasets/htrbao/aloha-dining-table/tree/main/videos

Read Example

from tsfile import TsFileReader

path = "data/aloha_dining_table.tsfile"
table = "aloha_dining_table"

with TsFileReader(path) as reader:
    columns = [
        "episode_index",
        "task_index",
        "frame_index",
        "sample_index",
        "action_0",
        "observation_state_0",
    ]
    with reader.query_table(table, columns, batch_size=4096) as result:
        batch = result.read_arrow_batch()
        print(batch)