eg.yaml

# ============================================================================
# Unified Metadata YAML Template for README.md and page assets.
# ============================================================================
# NOTE:
# - Canonical field names are the top-level keys in this file.
# - `alias` is DEPRECATED and kept only for backward compatibility.
# - External metadata should use canonical keys only.

# ========== HuggingFace Hub Configuration (README yaml)==========
# ---------- Start of the Yaml part of README ----------
# ------------------------------------------------------

task_categories:
  value: null
  required: true
  source: "fixed"
  alias: null
  default: "robotics"
# Task category labels, server as a label in huggingface hub.

language:
  value: null
  required: true
  source: "fixed"
  alias: null
  default: "en"
# Supported languages, server as a label in huggingface hub.

tags:
  value: null
  required: true
  source: "fixed"
  alias: null
  default:
    - "RoboCOIN"
    - "LeRobot"
# Dataset tags, server as a label in huggingface hub.

license:
  value: null
  required: true
  source: "fixed"
  alias:
    - "dataset_license"
  default: "apache-2.0"
# License type, server as a label in huggingface hub.

configs:
  value: null
  required: true
  source: "fixed"
  alias: null
  default: "default"
# Data file configuration

# ========== Gated Dataset Configuration ==========

extra_gated_prompt:
  value: null
  required: true
  source: "fixed"
  alias: null
  default: >-
    By accessing this dataset, you agree to cite the associated paper in your
    research/publications—see the "Citation" section for details. You agree to not
    use the dataset to conduct experiments that cause harm to human subjects.

extra_gated_fields:
  value: null
  required: true
  source: "fixed"
  alias: null
  default:
    Company/Organization:
      type: text
      description: e.g., "ETH Zurich", "Boston Dynamics", "Independent Researcher"
    Country:
      type: country
      description: e.g., "Germany", "China", "United States"
# Fields to be filled when accessing

# ---------- End of the Yaml part of README ----------
######################################################

# ========== Basic Information ==========

codebase_version:
  value: null
  required: true
  source: "fixed"
  alias: null
  default: "v2.1"
# that is codebase_version, get from local_dataset_info.yaml

dataset_name:
  value: null
  required: true
  source: "local_dataset_info.yaml"
  alias:
    - "task_name"
  default: "{robot_name}_{action}_{object}"
# Dataset Name, format: {robot}_{task_name}

dataset_uuid:
  value: null
  required: true
  source: "local_dataset_info.yaml"
  alias: null
  default: "00000000-0000-0000-0000-000000000000"
# Dataset UUID (usually auto-generated)

# ========== Environment Information ==========
scene_type:
  value: null
  required: true
  source: "local_dataset_info.yaml"
  alias:
    - "scene_level1"
    - "scene_level2"
    - "scene_level3"
    - "scene_level4"
    - "scene_level5"
    - "scene_level"
  default:
    level1: "scene_level1"
    level2: "scene_level2"
    level3: null
    level4: null
    level5: null
# Strictly fill in according to the scene library.
# --- README -----
# e.g. Catering-Restaurant.
# MUST follow the format: level1-level2-...
# ----------------

env_type:
  value: null
  required: true
  source: "local_dataset_info.yaml"
  alias: null
  default: "Due to some reasons, this dataset temporarily cannot provide the environment type information."
# --- README -----
# simulation/real.
# MUST follow the format: simulation/real_world. one in two choices
# ----------------

objects:
  value: null
  required: true
  source: "local_dataset_info.yaml"
  alias: null
  default:
    object_name: "object_name"
    color: "color"
    level1: "object"
    level2: "object"
    level3: null
    level4: null
    level5: null
# that is object_list, get from local_dataset_info.yaml
# --- README -----
# furniture-table.
# container-basket.
# MUST follow the format: object_name-level1-level2-...
# ----------------

# ========== Task Information ==========

task_operation_type:
  value: null
  required: true
  source: "local_dataset_info.yaml"
  alias: null
  default: "Due to some reasons, this dataset temporarily cannot provide the operation type information."
# dual_arm..., that is task_type.
# --- README -----
# simgle_arm/dual_arm/othes...
# MUST follow the format: simgle_arm/dual_arm/others... Choose one.
# ----------------

task_result:
  value: null
  # e.g.
  #   success/failure
  required: optional
  source: "annotations/subtask_annotations.jsonl"
  alias: null
  default: "null"
# --- README -----
# success/failure.
# MUST follow the format: success/failure. one in two choices
# ----------------

task_instruction:
  value: null
  # e.g.
  #   "put_apple_into_basket"
  required: true
  source: "local_dataset_info.yaml"
  alias:
    - "tasks"
    # Legacy compat: `pre_local_datasets_info.yaml` uses `task_descriptions`
    # (old field name) to store the main natural-language task instruction.
    # Purpose: map old `task_descriptions` -> standardized `task_instruction`.
    - "task_descriptions"
  default: "task_instruction"
# i.e. task_instruction, this is the natural language instruction.
# NOTE: MUST distinguish the task_instruction from task_descriptions.
# former is the natural language instruction, latter is the task sub-process description.(How to do the task)
# ----------------
# NOW, the task_instruction is the MAIN instruction in local_dataset_info.yaml,
# which is aligned to this yaml template. Here only use the first instruction as the MAIN instruction.
# bucause task_description is redundant with sub_tasks and deprecated, so we don't need to use it.
# ----------------

sub_tasks:
  value: null
  required: true
  source: "annotations/subtask_annotations.jsonl"
  alias:
    - "task_description"
  default:
    - "sub_task_1"
    - "sub_task_2"
    - "sub_task_3"
# that is subtask_list, get from annotations/subtask_annotations.jsonl
# NOTE: Is redundent with task_description, but here we keep it for compatibility.
# ONLY used as a reference for the task description.
# ----------------

atomic_actions:
  value: null
  required: true
  source: "local_dataset_info.yaml"
  alias: null
  default:
    - "atomic_action_1"
    - "atomic_action_2"
    - "atomic_action_3"
# that is so-called skill_list, get from local_dataset_info.yaml

# ========== Robot Configuration Info ==========

robot_name:
  value: null
  # e.g.
  #   "Agibot_g1"
  required: true
  source: "local_dataset_info.yaml"
  alias:
    - "device_model"
  default: "{manufacturer}+{robot_type}"
# "{Manifacurer}+{robot_type}"
# is device_model in local_dataset_info.yaml

end_effector_type:
  value: null
  # e.g.
  #   "two_finger_gripper"
  required: true
  source: "local_dataset_info.yaml"
  alias: null
  default: "end_effector_type"

tele_type:
  value: null
  # e.g.
  #   "master_arm"
  required: true
  source: "local_dataset_info.yaml"
  alias: null
  default: "Due to some reasons, this dataset temporarily cannot provide the teleoperation type information."

sensor_list:
  value: null
  required: true
  source: "meta/info.json"
  alias: null
  default: 
    - "sensor_1"
    - "sensor_2"
    - "sensor_3"
# Read from meta/info.json, from features.observation.<sensor_name,Iter items here.>
# usually have images and state.

came_info:
  value: null
  required: true
  source: "meta/info.json"
  alias:
    - "camera_info"
  default:
    observation.images.cam_name_1: "resolution_1"
    observation.images.cam_name_2: "resolution_2"
# CANNOT get from data providers usually. backward to resolution of the camera.

depth_enabled:
  value: null
  required: true
  source: "meta/info.json"
  alias: null
  default: "depth_enabled"
# Whether depth images are included (auto-calculated from features)

# ========== Coordinate Information ==========
########################################
# BELOW is to-be-checked if necessary. #
########################################

coordinate_definition:
  value: null
  required: true
  source: "fixed"
  alias: null
  default: "right-hand-frame"
# single str. directly display.

origin_xyz:
  value: null
  required: false
  source: "urdf"
  alias: null
  default: "origin_xyz"
# list of 3 floats. directly display.

#####################################
# UP is to-be-checked if necessary. #
#####################################

# ========== Dimension Information ==========

joint_rotation_dim:
  value: null
  required: true
  source: "meta/info.json"
  alias: null
  default: "joint_rotation_dim"
# choose from: radian, degree.
# single str. directly display.

end_rotation_dim:
  value: null
  required: true
  source: "meta/info.json"
  alias: null
  default: "end_rotation_dim"
# choose from: radian, degree.
# single str. directly display.
# that is the rotation dimension of the end effector.

end_translation_dim:
  value: null
  required: true
  source: "meta/info.json"
  alias: null
  default: "end_translation_dim"
# choose from: meter, millimeter.
# single str. directly display.
# that is the translation dimension of the end effector.

base_robtation_dim:
  value: null
  required: optional
  source: "meta/info.json"
  alias:
    - "base_rotation_dim"
  default: "base_rotation_dim"
# choose from: radian, degree.
# single str. directly display.
# that is the rotation dimension of the base of the robot.

base_translation_dim:
  value: null
  required: optional
  source: "meta/info.json"
  alias: null
  default: "base_translation_dim"
# choose from: meter, millimeter.
# single str. directly display.
# that is the translation dimension of the base of the robot.

# ========== Annotation Information ==========

annotations:
  value: null
  required: optional
  source: "annotations/"
  alias: null
  default:
    - "annotation_1"
    - "annotation_2"
    - "annotation_3"
# Iter through the file names in annotations/
# declaration of the annotations.

operation_platform_height:
  value: null
  required: false
  source: null
  alias: null
# Operation platform height (in centimeters, read from YAML file)
# TODO: Define the operation platform height usage.
# Temporarily keep it as-is.

# ========== Dataset Statistics ==========

statistics:
  value: null
  required: true
  source: "meta/episodes_stats.jsonl"
  alias: null
  default:
    total_episodes: "total_episodes"
    total_frames: "total_frames"
    total_tasks: "total_tasks"
    total_videos: "total_videos"
    total_chunks: "total_chunks"
    fps: "fps"
    dataset_size: "dataset_size"
    
# that is statistics, get from meta/info.json and calculation of file usage.

frame_num:
  value: null
  required: true
  source: "meta/info.json"
  alias: null
  default: "frame_num"

dataset_size:
  value: null
  required: true
  source: "auto"
  alias: null
  default: "dataset_size"
# Dataset file size (auto-calculated)

data_structure:
  value: null
  required: true
  source: "auto"
  alias:
    - "structure"
  default: "data_structure_str"
# that is the data structure of the dataset.
# value: str, the tree of the dataset.
# --- README -----
# display as-is.
# ----------------

# ========== Data Splits ==========

splits:
  value: null
  required: true
  source: "meta/info.json"
  alias: null
  default:
    train: "train"
# Data splits (read from meta/info.json)

# ========== File Path Patterns ==========
data_path:
  value: null
  required: false
  source: "auto"
  alias: null
  default: "data_path"
# Data file path pattern

video_path:
  value: null
  required: false
  source: "auto"
  alias: null
  default: "video_path"
# Video file path pattern

# ========== Feature Definitions ==========
features:
  value: null
  required: true
  source: "meta/info.json"
  alias: null
  default: 
    features_placeholder: "features_placeholder"
# Feature definitions (complete copy from meta/info.json)

# ========== Author and Link Information ==========

authors:
  value: null
  required: true
  source: "fixed"
  alias: null
  default:
    contributed_by:
      - name: "RoboCOIN Team at Beijing Academy of Artificial Intelligence (BAAI)"
# Author information

dataset_description:
  value: null
  required: true
  source: "fixed"
  alias: null
  default: "This dataset uses an extended format based on LeRobot and is fully compatible with LeRobot."

homepage:
  value: null
  required: true
  source: "fixed"
  alias: null
  default: "https://flagopen.github.io/RoboCOIN/"

paper:
  value: null
  required: true
  source: "fixed"
  alias: null
  default: "https://arxiv.org/abs/2511.17441"

repository:
  value: null
  required: true
  source: "fixed"
  alias: null
  default: "https://github.com/FlagOpen/RoboCOIN"

# ========== Contact and Support Information ==========
contact_detail:
  value: null
  required: false
  source: "fixed"
  alias:
    - "contact_email"
    - "dataset_email"
  default: "robocoin@baai.ac.cn"
# the contact detail: email, phone, etc.

contact_info:
  value: null
  required: true
  source: "fixed"
  alias: null
  default: "For questions, issues, or feedback regarding this dataset, please contact us."
# the contact info: explanation str.

support_info:
  value: null
  required: true
  source: "fixed"
  alias: null
  default: "For technical support, please open an issue on our GitHub repository."
# the support info: explanation str.

# ========== License Details ==========
license_details:
  value: null
  required: true
  source: "fixed"
  alias: null
  default: "apache-2.0"

# ========== Citation Information ==========

citation_bibtex:
  value: null
  required: true
  source: "fixed"
  alias: null
  default: |
    @article{robocoin,
      title={RoboCOIN: An Open-Sourced Bimanual Robotic Data Collection for Integrated Manipulation},
      author={Shihan Wu, Xuecheng Liu, Shaoxuan Xie, Pengwei Wang, Xinghang Li, Bowen Yang, Zhe Li, Kai Zhu, Hongyu Wu, Yiheng Liu, Zhaoye Long, Yue Wang, Chong Liu, Dihan Wang, Ziqiang Ni, Xiang Yang, You Liu, Ruoxuan Feng, Runtian Xu, Lei Zhang, Denghang Huang, Chenghao Jin, Anlan Yin, Xinlong Wang, Zhenguo Sun, Junkai Zhao, Mengfei Du, Mingyu Cao, Xiansheng Chen, Hongyang Cheng, Xiaojie Zhang, Yankai Fu, Ning Chen, Cheng Chi, Sixiang Chen, Huaihai Lyu, Xiaoshuai Hao, Yequan Wang, Bo Lei, Dong Liu, Xi Yang, Yance Jiao, Tengfei Pan, Yunyan Zhang, Songjing Wang, Ziqian Zhang, Xu Liu, Ji Zhang, Caowei Meng, Zhizheng Zhang, Jiyang Gao, Song Wang, Xiaokun Leng, Zhiqiang Xie, Zhenzhen Zhou, Peng Huang, Wu Yang, Yandong Guo, Yichao Zhu, Suibing Zheng, Hao Cheng, Xinmin Ding, Yang Yue, Huanqian Wang, Chi Chen, Jingrui Pang, YuXi Qian, Haoran Geng, Lianli Gao, Haiyuan Li, Bin Fang, Gao Huang, Yaodong Yang, Hao Dong, He Wang, Hang Zhao, Yadong Mu, Di Hu, Hao Zhao, Tiejun Huang, Shanghang Zhang, Yonghua Lin, Zhongyuan Wang and Guocai Yao},
      journal={arXiv preprint arXiv:2511.17441},
      url = {https://arxiv.org/abs/2511.17441},
      year={2025},
      }

additional_citations:
  value: null
  required: true
  source: "fixed"
  alias: null
  default: |
    If you use this dataset, please also consider citing:
    LeRobot Framework: https://github.com/huggingface/lerobot

# ========== Version Information ==========
version_info:
  value: null
  required: true
  source: "fixed"
  alias: null
  default: "Initial Release"

# ========== Auto-generated Fields (usually no manual input required) ==========
# The following fields are usually auto-generated by the system, for reference only

# REMOVED FIELDS (no longer needed):
# - path: Dataset path identifier (auto-generated)
# - thumbnail_url: Thumbnail URL (auto-generated)

video_url:
  value: null
  required: optional
  source: "auto"
  alias: null
  default: null
# Relative path to one preview video (high/head/front camera), for README/Hub. Auto-filled via _vid_coll in apply_auto_fields.

# ========== Raw Data Retention ==========
raw:
  value: null
  required: false
  source: null
  alias: null
# Complete object after original YAML parsing (for debugging and extension)

# ========== Deactivated ==========
# The following fields are defined but are not currently active in this dataset configuration.

# Removed deprecated duplicated label field:
# - segment_level_label
# Keep this comment only for migration context.

surface_material:
  value: null
  required: false
  source: null
  alias: null
# Ground surface material, such as concrete, wood, tile, grass, etc.

light:
  value: null
  required: false
  source: null
  alias: null
# Light source information: light intensity, color temperature attributes, shadowless lamp.

noise_level:
  value: null
  required: false
  source: null
  alias: null
# Environmental noise level in decibels (dB).

tactile_info:
  value: null
  required: false
  source: null
  alias: null
# Tactile sensor information: manufacturer, model, sensing method, etc.

sound_info:
  value: null
  required: false
  source: null
  alias: null
# Sound sensor information: manufacturer, model, sensing method, etc.

base_transform:
  value: null
  required: false
  source: null
  alias: null
# Spatial transformation relationship of the body coordinate system relative to the base coordinate system.

end_transform:
  value: null
  required: false
  source: null
  alias: null
# Spatial transformation relationship of the end effector coordinate system relative to the base coordinate system.

# Removed deprecated duplicated label fields:
# - Trajectory-level_label
# - Segment-level_label
# - Frame-level_label
# Keep this comment only for migration context.

data_collector:
  value: null
  required: false
  source: null
  alias: null
# Name of the data collector.
# CANNOT get from data providers usually. thus set to false.

data_annotator:
  value: null
  required: optional
  source: null
  alias: null
# Name of the data annotator.
# CANNOT get from data providers usually. thus set to false.

data_modifier:
  value: null
  required: false
  source: null
  alias: null
# Name of the data modifier.

modify_logo:
  value: null
  required: false
  source: null
  alias: null
# Record of dataset modification information.

data_dowload:
  value: null
  required: false
  source: null
  alias: null
# Dataset download method description.

field_definiton:
  value: null
  required: false
  source: null
  alias: null
# Standard field definition description.

dataset_organization:
  value: null
  required: false
  source: null
  alias: null
# Submitting organization, e.g., BAAI.

dataset_email:
  value: null
  required: false
  source: null
  alias: null
# Contact email, e.g., robocoin@baai.ac.cn.

dataset_license:
  value: null
  required: false
  source: null
  alias: null
# Open source license type: Apache License 2.0, CC BY 4.0, MIT license, Unsure.

# ========== Deprecated Fields ==========
# Removed deprecated duplicated canonical fields:
# - task_description (use canonical `sub_tasks`, alias compatibility only)
# - task_name (use canonical `dataset_name`, alias compatibility only)