Dataset Viewer
The dataset viewer is not available because its heuristics could not detect any supported data files. You can try uploading some data files, or configuring the data files location manually.
🤖 humanoid_long_horizon_tasks 人形机器人长周期任务数据集
本仓库是人形机器人长程任务数据集,用于记录和共享人形机器人在真实/仿真环境中完成复杂长周期任务的演示数据,为机器人学习、强化学习、具身智能研究提供高质量训练数据。
📋 任务列表(持续更新中)
| 任务名称 | 数据文件 | 状态 | 更新时间 | 说明 |
|---|---|---|---|---|
pick_water_bottle_handover_and_insert |
water_bottle_pick_and_insert.rar |
✅ 已上传 | 2026-05-11 | 水瓶抓取-递接-插入任务演示数据 |
| (后续任务1) | (对应压缩包文件名) | ⏳ 待上传 | - | (任务说明) |
| (后续任务2) | (对应压缩包文件名) | ⏳ 待上传 | - | (任务说明) |
📂 数据集结构说明
所有任务数据以 .rar 压缩包形式存储,下载后解压即可使用,统一结构如下:
water_bottle_pick_and_insert/
├── episode_0030/ # 单轮任务演示数据
│ ├── colors/ # RGB相机图像序列
│ │ ├── 000000_color_0.jpg
│ │ ├── 000000_color_1.jpg
│ │ └── ...
│ └── data.json # 任务状态、动作序列、时间戳等元数据
├── episode_0031/
└── ...
episode_xxxx/:单轮完整任务的演示数据,每一个文件夹对应一次从任务开始到结束的完整流程colors/:多视角相机采集的RGB图像序列,按时间戳命名data.json:包含任务过程中的机器人关节状态、末端位姿、动作指令、时间戳等结构化数据
🛠️ 使用指南
1. 下载数据集
方式1:直接在仓库页面点击 .rar 文件右侧的下载按钮
方式2:使用 huggingface_hub Python API 下载(推荐)
from huggingface_hub import hf_hub_download
# 下载水瓶抓取任务数据
file_path = hf_hub_download(
repo_id="yky666/humanoid_long_horizon_tasks",
filename="water_bottle_pick_and_insert.rar",
repo_type="dataset"
)
print(f"文件下载路径:{file_path}")
2. 解压数据
使用WinRAR/7-Zip等工具解压 .rar 压缩包,即可得到完整的 episode 文件夹结构。
3. 读取数据示例
import json
import os
from PIL import Image
# 读取单轮任务数据
episode_dir = "./water_bottle_pick_and_insert/episode_0030"
data_path = os.path.join(episode_dir, "data.json")
# 读取结构化元数据
with open(data_path, "r", encoding="utf-8") as f:
task_data = json.load(f)
print(f"任务总帧数:{len(task_data['timestamps'])}")
# 读取图像序列
colors_dir = os.path.join(episode_dir, "colors")
image_files = sorted(os.listdir(colors_dir))
for img_file in image_files[:5]: # 读取前5帧
img = Image.open(os.path.join(colors_dir, img_file))
print(f"加载图像:{img_file},尺寸:{img.size}")
4. 项目部署
可参照 https://github.com/yky666/humanoid_robot_long_horizon_tasks.git
📌 后续更新计划
- 持续上传更多人形机器人长周期操作任务数据
- 补充任务说明文档、数据标注规范
- 提供配套的数据预处理脚本与可视化工具
📄 许可证
本数据集基于 Apache-2.0 许可证开源,欢迎用于学术研究与非商业用途,二次分发请注明来源。
- Downloads last month
- 67