---
title: Linkerhand-Open-World-Dataset
canonical_url: "https://www.modelscope.cn/datasets/Linkerbot/Linkerhand-Open-World-Dataset"
md_url: "https://www.modelscope.cn/datasets/Linkerbot/Linkerhand-Open-World-Dataset.md"
repository: Linkerbot/Linkerhand-Open-World-Dataset
last_updated: 2026-10-12
license: "Apache License 2.0"
storage_size: "6.1 GB"
downloads: 3023
stars: 2
---

# Linkerhand-Open-World-Dataset

> Linkerhand-Open-World-Dataset - Linkerbot 在 ModelScope 开源的数据集。Open-source dataset collected with Linkerhand dexterous hands exploring the world and completing human-like tasks.

Linkerbot/Linkerhand-Open-World-Dataset 是 ModelScope 魔搭社区上的数据集，存储大小 6.1 GB，采用 Apache License 2.0 许可。

- **Repository**: Linkerbot/Linkerhand-Open-World-Dataset
- **License**: Apache License 2.0
- **Storage size**: 6.1 GB
- **Downloads**: 3023
- **Stars**: 2
- **Last updated**: 2026-10-12

Source: https://www.modelscope.cn/datasets/Linkerbot/Linkerhand-Open-World-Dataset

---

## Linkerhand Open-World Dataset

The Linkerhand Open-World Dataset provides large-scale, high-quality robot manipulation data to accelerate research in perception, control, and generalization. Our pick-and-place scenes capture diverse objects, viewpoints, and motion patterns, enabling training and evaluation of robust policies and multimodal models. It targets LeRobot open-source dataset creation and model training.

### Current contents
- Over 15 hours of synchronized robot and vision data from pick-and-place scenes to human-like manipulation scenes
- Current objects include: real apples, building blocks, cabbage toy, magic cubes, Miduoqi bread snacks, purple doll, and tennis balls
- We are continuously adding more scenes and data to expand coverage and diversity

### Directory structure
```text
pick-and-place/
└── tasks_names/
    └── session_timestamp/
        ├── episode_000001/
        │   ├── telemetry.npz                 # Hands + Arms Observations and Actions
        │   ├── cameras/
        │   │   ├── cam_top.mp4               # H.264 compressed video
        │   │   └── cam_top.timestamps.npy    # Frame timestamps
        │   ├── camera_info.json              # Camera intrinsics
        │   ├── manifest.json                 # Episode metadata
        │   └── metadata.json                 # Episode statistics
        └── ... (same structure)
```

For dataset file metadata and data files, please visit the "Files and versions" tab. 

#### DOWNLOAD METHOD 
:modelscope-code[]{type="sdk"}
:modelscope-code[]{type="git"}