CollisionCode/RoboCerebra_lerobot_v2.1
收藏Hugging Face2026-04-07 更新2026-04-12 收录
下载链接:
https://hf-mirror.com/datasets/CollisionCode/RoboCerebra_lerobot_v2.1
下载链接
链接失效反馈官方服务:
资源简介:
---
license: apache-2.0
task_categories:
- robotics
tags:
- libero
- lerobot
- embodied-ai
- manipulation
pretty_name: RoboCerebra (LeRobot v2.1 Format)
size_categories:
- 1G<n<10G
---
# RoboCerebra (LeRobot v2.1 Format)
## Dataset Description
This dataset is a converted version of [**RoboCerebra**](https://github.com/qiuboxiang/RoboCerebra), now formatted for compatibility with the **[LeRobot](https://github.com/huggingface/lerobot)** ecosystem (v2.1).
While the official RoboCerebra dataset only provides HDF5 files and RLDS conversion methods, this version enables seamless training with VLA models within the LeRobot framework.
- **Original Dataset:** [RoboCerebra](https://github.com/qiuboxiang/RoboCerebra)
- **Conversion Tool:** [any4lerobot](https://github.com/Tavish9/any4lerobot)
- **Format:** LeRobot v2.1
- **Robot:** Panda Robot (7-DOF Arm + Gripper)
- **Tasks:** 6 categories of long-horizon memory-based manipulation tasks
## Dataset Structure
The dataset follows the standard LeRobot v2.1 structure:
- **`data/`**: Contains episode data in Parquet format.
- **`meta/`**: Contains task definitions, statistics, and camera configurations.
- **`videos/`**: (If included) High-quality task execution videos.
### Key Observations/Actions
- `observation.images.agentview`: 256x256 agent-view camera RGB.
- `observation.images.eye_in_hand`: 256x256 wrist-mounted camera RGB.
- `observation.state`: Robot joint positions and gripper state.
- `action`: 7-DOF delta end-effector pose + gripper action.
许可证:Apache-2.0
任务类别:
- 机器人学
标签:
- libero
- LeRobot
- 具身人工智能(Embodied AI)
- 机器人操作
友好名称:RoboCerebra(LeRobot v2.1格式版)
规模类别:1G<n<10G
# RoboCerebra(LeRobot v2.1格式版)
## 数据集描述
本数据集是[**RoboCerebra**](https://github.com/qiuboxiang/RoboCerebra)的转换版本,现已适配**[LeRobot](https://github.com/huggingface/lerobot)**生态系统(v2.1版本)的格式要求。
官方原版RoboCerebra数据集仅提供HDF5文件与RLDS转换方法,而本版本可在LeRobot框架中无缝适配VLA模型开展训练。
- **原始数据集**:[RoboCerebra](https://github.com/qiuboxiang/RoboCerebra)
- **转换工具**:[any4lerobot](https://github.com/Tavish9/any4lerobot)
- **格式**:LeRobot v2.1
- **机器人平台**:潘达(Panda)机器人(7自由度机械臂+夹爪)
- **任务类型**:6类基于长时序记忆的机器人操作任务
## 数据集结构
本数据集遵循标准LeRobot v2.1结构:
- **`data/`**:存储Parquet格式的任务回合数据。
- **`meta/`**:包含任务定义、统计信息与相机配置参数。
- **`videos/`**:(若包含)高质量任务执行视频。
### 关键观测与动作
- `observation.images.agentview`:256×256分辨率的智能主体视角RGB图像
- `observation.images.eye_in_hand`:256×256分辨率的腕部安装相机RGB图像
- `observation.state`:机器人关节位置与夹爪状态
- `action`:7自由度末端执行器增量位姿+夹爪动作
提供机构:
CollisionCode



