rate70_edge_54

This dataset converts iiyudana/rate70_edge_54 from LeRobot v2.1 Parquet frames to Apache TsFile. It records an ALOHA robot task: pick up a cube with the right arm and transfer it to the left arm.

• Original dataset: iiyudana/rate70_edge_54
• License: Apache-2.0
• Robot type: ALOHA
• Split: train
• Scale: 54 episodes, 21,600 frames, 50 fps, 1 task, 1 TsFile
• Modalities: Time-series

Files

• data/rate70_edge_54.tsfile: numeric robot time-series data.
• meta/: source metadata mirrored with meta/info.json rewritten for the TsFile layout.
• Videos are not included. They remain in the source dataset videos tree.

Schema

The TsFile table is named rate70_edge_54.

Role	Columns
Time	Time, INT64 milliseconds
TAG	episode_index, task_index
FIELD	frame_index, sample_index
FIELD	observation_state_0 through observation_state_13, FLOAT
FIELD	action_0 through action_13, FLOAT

The original 14-element observation.state and action vectors are flattened into scalar FLOAT measurements. 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.
• All 21,600 numeric observations are retained. No source rows or numeric measurements are intentionally dropped apart from this redundant timestamp representation.
• The original top-camera MP4 video streams are excluded from this conversion and remain available from the source dataset.

Read example

Read the uploaded file with an Apache TsFile SDK:

from tsfile import TsFileReader

with TsFileReader("data/rate70_edge_54.tsfile") as reader:
    tables = reader.get_all_table_schemas()
    print(tables["rate70_edge_54"])