libero_object_mask_depth_ipec_community_format

LIBERO Object Mask Depth (IPEC Community Format) (TsFile)

Apache TsFile version of binhng/libero_object_mask_depth_IPEC_COMMUNITY_format.

Overview

A Franka robot dataset for the LIBERO-Object manipulation suite, distributed in the IPEC community format with object-mask and depth observations. Each of the ten tasks asks the robot to pick up a household object and place it in the basket. The converted time series records the robot's end-effector and joint state together with the commanded action at every control step.

• Scale: 500 episodes, 74,507 frames, 10 tasks, sampled at 20 fps; one train split in data/libero_object_mask_depth_ipec_community_format.tsfile.

Tasks (task_index): 0 pick up the salad dressing and place it in the basket; 1 bbq sauce; 2 ketchup; 3 tomato sauce; 4 alphabet soup; 5 cream cheese; 6 chocolate pudding; 7 butter; 8 orange juice; 9 milk.

Schema (TsFile structure)

• Time (INT64, milliseconds) - per-episode sample time, round(timestamp * 1000) in milliseconds (the source timestamp column equals Time / 1000 seconds and is therefore not stored separately).
• episode_index, task_index (TAG) - device dimensions identifying one demonstration episode and one task; query a single episode with WHERE episode_index=0.
• observation_state_0..7, observation_states_ee_state_0..5, observation_states_joint_state_0..6, observation_states_gripper_state_0..1, action_0..6 (FIELD) - FLOAT robot joint/end-effector/gripper state and action measurements (8-D state, 6-D ee_state, 7-D joint_state, 2-D gripper_state, 7-D action).
• frame_index, sample_index (INT64, FIELD) - source frame counter and source row counter (index renamed to sample_index).

The dataset is stored as one wide TsFile table; within the train split all episodes share that single file with episode_index and task_index as TAG columns. State/action vectors are flattened to one FLOAT field per element (. -> _, element index appended). Camera video files are not included in this repository; the original dataset's videos are at videos/. meta/ is mirrored from the source. No data rows or non-video columns are dropped, apart from the redundant source timestamp column (replaced by Time). Eight camera image/video features (observation.images.*) are omitted as non-time-series.

Usage

Install the Apache TsFile Python SDK (pip install tsfile) and read a converted file:

from pathlib import Path
from tsfile import TsFileReader

path = Path("data/libero_object_mask_depth_ipec_community_format.tsfile")
with TsFileReader(str(path)) as reader:
    schemas = reader.get_all_table_schemas()
    print("tables:", list(schemas))
    table_name = next(iter(schemas))
    table = schemas[table_name]
    columns = [column.get_column_name() for column in table.get_columns()]
    print("columns:", columns)
    field_names = [
        column.get_column_name()
        for column in table.get_columns()
        if column.get_column_name() not in {"Time", "time"}
    ]
    if field_names:
        with reader.query_table(table_name, field_names[:3], batch_size=1024) as result:
            batch = result.read_arrow_batch()
            if batch is not None:
                print(batch.to_pandas().head())

Source & license

• Original dataset: https://huggingface.co/datasets/binhng/libero_object_mask_depth_IPEC_COMMUNITY_format
• Author / publisher: binhng
• License: apache-2.0