droid_subsets_scooping (TsFile)

Apache TsFile version of jellyho/droid_subsets_scooping.

Overview

A LeRobot v3.0 manipulation dataset recorded with a Franka robot arm, converted to a single Apache TsFile table.

• Scale: 300 episodes / 147,394 frames at 15 fps (~164 min of robot time) / 300 task(s).

• Task language: Move the black pot to the right then use the spoon to take some contents out of the pot; Scoop up the beans from the bowl on the right and put them to the bowl on the left; Use the silver spoon to get some coffee beans from the container and put them on the small white plate.; Use the spoon to scoop some of the contents in the white bowl into the glass on the right; Use the silver spoon to scoop some of the contents in the bowl into the glass cup; Use the spoon to transfer beans to the bowl on the left; Use the spoon to scoop some beans from the red bowl and put them in the pot; Use the spoon to move some of the brown objects to the other bowl …

• Tables: 1 TsFile table(s), 147,394 rows in total.

  • droid_subsets_scooping: 147,394 rows, Time restarts inside every episode.

Schema (TsFile structure)

droid_subsets_scooping.tsfile

• Time (INT64, milliseconds) — round(timestamp × 1000); the source timestamp column is folded into Time.
• TAG: episode_index, task_index — device dimension; query one device with WHERE episode_index=....
• FIELD (98): frame_index (int64), sample_index (int64), is_first (int64), is_last (int64), is_terminal (int64), language_instruction (string), language_instruction_2 (string), language_instruction_3 (string), observation_state_gripper_position (float), action_gripper_position (float), action_gripper_velocity (float), discount (float), reward (float), task_category (string), building (string), collector_id (string), date (string), is_episode_successful (int64), observation_state_cartesian_position_0 (float), observation_state_cartesian_position_1 (float), observation_state_cartesian_position_2 (float), observation_state_cartesian_position_3 (float), observation_state_cartesian_position_4 (float), observation_state_cartesian_position_5 (float), observation_state_joint_position_0 (float), observation_state_joint_position_1 (float), observation_state_joint_position_2 (float), observation_state_joint_position_3 (float), observation_state_joint_position_4 (float), observation_state_joint_position_5 (float), observation_state_joint_position_6 (float), observation_state_0 (float), observation_state_1 (float), observation_state_2 (float), observation_state_3 (float), observation_state_4 (float), observation_state_5 (float), observation_state_6 (float), observation_state_7 (float), action_cartesian_position_0 (float), action_cartesian_position_1 (float), action_cartesian_position_2 (float), action_cartesian_position_3 (float), action_cartesian_position_4 (float), action_cartesian_position_5 (float), action_cartesian_velocity_0 (float), action_cartesian_velocity_1 (float), action_cartesian_velocity_2 (float), action_cartesian_velocity_3 (float), action_cartesian_velocity_4 (float), action_cartesian_velocity_5 (float), action_joint_position_0 (float), action_joint_position_1 (float), action_joint_position_2 (float), action_joint_position_3 (float), action_joint_position_4 (float), action_joint_position_5 (float), action_joint_position_6 (float), action_joint_velocity_0 (float), action_joint_velocity_1 (float), action_joint_velocity_2 (float), action_joint_velocity_3 (float), action_joint_velocity_4 (float), action_joint_velocity_5 (float), action_joint_velocity_6 (float), action_original_0 (float), action_original_1 (float), action_original_2 (float), action_original_3 (float), action_original_4 (float), action_original_5 (float), action_original_6 (float), action_0 (float), action_1 (float), action_2 (float), action_3 (float), action_4 (float), action_5 (float), action_6 (float), action_7 (float), camera_extrinsics_wrist_left_0 (float), camera_extrinsics_wrist_left_1 (float), camera_extrinsics_wrist_left_2 (float), camera_extrinsics_wrist_left_3 (float), camera_extrinsics_wrist_left_4 (float), camera_extrinsics_wrist_left_5 (float), camera_extrinsics_exterior_1_left_0 (float), camera_extrinsics_exterior_1_left_1 (float), camera_extrinsics_exterior_1_left_2 (float), camera_extrinsics_exterior_1_left_3 (float), camera_extrinsics_exterior_1_left_4 (float), camera_extrinsics_exterior_1_left_5 (float), camera_extrinsics_exterior_2_left_0 (float), camera_extrinsics_exterior_2_left_1 (float), camera_extrinsics_exterior_2_left_2 (float), camera_extrinsics_exterior_2_left_3 (float), camera_extrinsics_exterior_2_left_4 (float), camera_extrinsics_exterior_2_left_5 (float)

Conversion notes

• LeRobot v3.0 robot dataset converted with the project's generic lerobot converter: all episodes of the source train split share one TsFile, with the source columns episode_index and task_index as TAG columns (the TsFile device dimension). Query a single episode with WHERE episode_index=0.
• Vector columns preserve the full source column name when flattened: . becomes _ and the element index is appended, so observation.state becomes observation_state_0..N-1 and action becomes action_0..N-1, all single-precision FLOAT. frame_index is kept and the source index column is kept as sample_index.
• Time = round(timestamp * 1000) ms and restarts inside each episode. The source timestamp column is not stored separately because it equals Time / 1000 seconds.
• Camera video streams are NOT included in this repository; see the original dataset for them: https://huggingface.co/datasets/jellyho/droid_subsets_scooping/tree/main/videos
• meta/ is mirrored from the source with meta/info.json updated to describe the converted TsFile artifact. No numeric time-series row is dropped.
• The source is a merged DROID skill subset with 300 distinct language tasks; the language instructions stay in meta/tasks.parquet and are referenced by task_index, which is a TAG here.

Usage

Read this dataset with the Java SDK. It is written with the zstd compressor. The Apache TsFile Java SDK reads every compressor, but the Python SDK (pip install tsfile) implements only UNCOMPRESSED, SNAPPY, GZIP and LZ4; opening this file with it yields no rows.

Read a converted file with the Apache TsFile Java SDK (org.apache.tsfile:tsfile):

import org.apache.tsfile.file.metadata.IDeviceID;
import org.apache.tsfile.read.TsFileReader;
import org.apache.tsfile.read.TsFileSequenceReader;
import org.apache.tsfile.read.common.Path;
import org.apache.tsfile.read.expression.QueryExpression;
import org.apache.tsfile.read.query.dataset.QueryDataSet;
import org.apache.tsfile.write.schema.MeasurementSchema;

import java.util.ArrayList;
import java.util.List;

String path = "data/droid_subsets_scooping.tsfile";
try (TsFileSequenceReader sequenceReader = new TsFileSequenceReader(path);
     TsFileReader reader = new TsFileReader(sequenceReader)) {
    // Devices are enumerated as IDeviceID objects. Tables with TAG columns
    // have multi-segment device ids that cannot be rebuilt from their string
    // form, so always use getAllDevices() rather than getAllDeviceIds().
    for (IDeviceID device : sequenceReader.getAllDevices()) {
        List<Path> paths = new ArrayList<>();
        for (MeasurementSchema schema : reader.getMeasurement(device)) {
            String name = schema.getMeasurementName();
            if (name != null && !name.isEmpty() && !name.equalsIgnoreCase("Time")) {
                paths.add(new Path(device, name, true));
            }
        }
        QueryDataSet dataSet = reader.query(QueryExpression.create(paths, null));
        while (dataSet.hasNext()) {
            System.out.println(dataSet.next());
        }
    }
}

Source & license

• Original dataset: jellyho/droid_subsets_scooping
• License: apache-2.0 (as declared by the original dataset).