Datasets:

THULab
/

soarm100_data2

license: apache-2.0
task_categories:
  - robotics
tags:
  - LeRobot
  - so100
  - tutorial
  - tsfile
  - format:tsfile
  - time-series
  - timeseries
modality:
  - tabular
  - timeseries
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/soarm100_data2.tsfile
size_categories:
  - 10K<n<100K

soarm100_data2 (TsFile)

Apache TsFile version of Dangvi/soarm100_data2.

Overview

This dataset is a LeRobot v2.1 dataset of an SO-100 arm.

Robot: SO-100 (so100)
Scale: 75 episodes, 33,440 frames, 30 fps
Split: train
Cameras (not uploaded): wristcam, webcam

Schema (TsFile structure)

The TsFile table is named soarm100_data2.

Role	Columns
Time	`Time`, INT64 milliseconds
TAG	`episode_index`, `task_index`
FIELD	`frame_index`, `sample_index`
FIELD	`action_0` … `action_5`, FLOAT
FIELD	`observation_state_0` … `observation_state_5`, FLOAT

Vector columns are flattened into scalar FLOAT measurements (single precision). The source index column is retained as sample_index.

Conversion notes

Time = round(timestamp * 1000) with millisecond precision; time restarts inside each episode, while episode_index and task_index identify the TsFile device.
The original timestamp field is omitted because it is exactly represented by Time / 1000.
Camera video streams are NOT included in this repository; they remain in the source videos tree.
meta/ is mirrored from the source. Aside from the redundant timestamp column and the excluded videos, no source rows or numeric fields are dropped.

Read example

from tsfile import TsFileReader

with TsFileReader("data/soarm100_data2.tsfile") as reader:
    print(reader.get_all_table_schemas()["soarm100_data2"])

Source & license

Original dataset: https://huggingface.co/datasets/Dangvi/soarm100_data2
Author: Dangvi
License: apache-2.0