MuJoCoPickLift
收藏资源简介:
该数据集是一个用于机器人任务的数据集,专门针对MuJoCoPickLift任务。它包含5个episodes,总计2336帧,使用so101类型机器人。数据以30fps的帧率采集,包括动作(6维浮点数组,对应机器人关节位置)、观测状态(6维浮点数组,同样表示关节位置)、观测图像(来自手腕和头顶视角的480x640 RGB视频)、奖励(单值浮点)、时间戳、帧索引、episode索引等特征。数据集结构以parquet文件格式存储,视频文件为mp4格式,适用于机器人控制和强化学习研究。
This dataset is tailored for robotic tasks, with a specific focus on the MuJoCoPickLift task. It contains 5 episodes totaling 2336 frames, and uses the so101 robot. The data was collected at a frame rate of 30 fps, and includes the following features: actions (6-dimensional float arrays corresponding to robot joint positions), observation states (6-dimensional float arrays that also represent joint positions), observation images (480×640 RGB videos captured from wrist and top-down viewpoints), rewards (single-value floating-point numbers), timestamps, frame indices, episode indices, and other relevant attributes. The dataset is stored in Parquet file format, while the video files are saved in MP4 format. This dataset is applicable to research in robotic control and reinforcement learning.
数据集概述
数据集名称: MuJoCoPickLift
数据集地址: https://huggingface.co/datasets/johnsutor/MuJoCoPickLift
许可协议: Apache-2.0
任务类别: 机器人学 (Robotics)
标签: LeRobot
数据集基本信息
| 属性 | 值 |
|---|---|
| 机器人类型 | so101 |
| 总片段数 (Episodes) | 5 |
| 总帧数 (Frames) | 2336 |
| 总任务数 (Tasks) | 1 |
| 数据块大小 (Chunks Size) | 1000 |
| 数据文件大小 | 100 MB |
| 视频文件大小 | 200 MB |
| 帧率 (FPS) | 30 |
| 训练/测试划分 | 全部用于训练 (split: train, 0:5) |
数据集结构
- 数据格式: Parquet 文件
- 视频格式: MP4 文件 (AV1 编码)
- 数据路径:
data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet - 视频路径:
videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4
特征字段
| 字段 | 数据类型 | 形状 | 描述 |
|---|---|---|---|
| action | float32 | [6] | 机器人动作:肩部旋转、肩部升降、肘部弯曲、腕部弯曲、腕部旋转、夹爪位置 |
| observation.state | float32 | [6] | 机器人状态(与动作相同维度):肩部旋转、肩部升降、肘部弯曲、腕部弯曲、腕部旋转、夹爪位置 |
| observation.images.wrist | video | [480, 640, 3] | 腕部摄像头图像 (480x640, RGB, AV1 编码, 30fps) |
| observation.images.overhead | video | [480, 640, 3] | 顶部摄像头图像 (480x640, RGB, AV1 编码, 30fps) |
| reward | float32 | [1] | 奖励值 |
| timestamp | float32 | [1] | 时间戳 |
| frame_index | int64 | [1] | 帧索引 |
| episode_index | int64 | [1] | 片段索引 |
| index | int64 | [1] | 全局索引 |
| task_index | int64 | [1] | 任务索引 |



