README.md

metadata

license: mit
task_categories:
  - robotics
language:
  - en
tags:
  - robotics
  - lerobot
  - manipulation
  - bimanual
size_categories:
  - 1K<n<10K
pretty_name: Bone to Pick Combined

Bone to Pick Combined

This is a LeRobot v2.1 format dataset containing bimanual robot manipulation demonstrations.

Dataset Structure

• Episodes: 20 episodes of robot manipulation
• Total Frames: 7,788 frames
• Cameras: 3 camera views per episode
  • observation.images.my_blue_follower_arm_front
  • observation.images.my_white_follower_arm_front
  • observation.images.my_white_follower_arm_global_top
• Robot: Bimanual manipulator (12 joints)
• Task: "Grasp the blue bone and place it inside of the box"

Usage with LeRobot

from lerobot.common.datasets.lerobot_dataset import LeRobotDataset

dataset = LeRobotDataset("Sraghvi/Bone_to_Pick_Combined")

Dataset Format

This dataset follows the LeRobot v2.1 format:

• data/chunk-000/: Parquet files containing episode data
• videos/chunk-000/: MP4 video files for each camera view
• meta/: Metadata files
  • info.json: Dataset configuration and feature descriptions
  • episodes/chunk-000/file-000.parquet: Episode metadata
  • tasks.parquet: Task descriptions
  • stats.json: Dataset statistics

Features

• Action: 12-dimensional (6 joints per arm + 2 grippers)
• Observation State: 12-dimensional robot state
• Images: 3 camera views (480x640x3)
• Timestamp: Frame timestamps
• Episode Index: Episode identifier
• Frame Index: Frame number within episode
• Task Index: Task identifier

Citation

Collected as part of the Bone to Pick bimanual manipulation project.