AkshitMajorProjectMIR1_combined
收藏资源简介:
该数据集是一个由LeRobot创建的机器人操作任务数据集,用于训练机器人控制模型。它包含机器人执行任务时的动作和观测数据:动作数据包括6个关节位置(肩部平移、肩部提升、肘部弯曲、手腕弯曲、手腕旋转和夹爪位置),观测状态同样包含这6个关节位置,并额外提供两个视频源:顶部摄像头和手腕摄像头。视频规格为分辨率480x640、帧率30fps、编码为av1。数据集总共有42个episodes,包含28057帧,数据文件大小为100MB,视频文件大小为200MB,机器人类型为so_follower。数据以parquet格式存储,便于高效处理和模型训练。
This dataset is a robotic manipulation task dataset created by LeRobot, designed for training robotic control models. It contains action and observation data collected during robot task execution: the action data includes 6 joint positions (shoulder translation, shoulder lift, elbow flexion, wrist flexion, wrist rotation, and gripper position), while the observation data also includes these 6 joint positions, and additionally provides two video sources: the top camera and the wrist-mounted camera. The videos have a resolution of 480×640, a frame rate of 30 fps, and use the AV1 encoding format. In total, the dataset contains 42 episodes, 28057 frames, with a data file size of 100 MB and a video file size of 200 MB. The robotic platform used is so_follower. The data is stored in Parquet format to facilitate efficient processing and model training.
数据集概述
- 数据集名称:AkshitMajorProjectMIR1_combined
- 许可证:Apache-2.0
- 任务类别:机器人学(Robotics)
- 标签:LeRobot
- 创建工具:LeRobot
数据集规模
- 总片段数(episodes):42
- 总帧数(frames):28057
- 总任务数(tasks):1
- 数据文件大小:约100 MB
- 视频文件大小:约200 MB
- 帧率(FPS):30
- 代码库版本:v3.0
数据集结构
- 配置:仅有一个默认配置 (
default),数据文件路径为data/*/*.parquet。 - 数据拆分:训练集 (
train) 包含全部42个片段。
特征(Features)
| 特征名 | 数据类型 | 形状 | 描述 |
|---|---|---|---|
action |
float32 | [6] | 6维动作指令,包括肩部旋转、肘部弯曲、腕部弯曲/旋转和夹爪位置 |
observation.state |
float32 | [6] | 机器人状态观测,维度与动作一致 |
observation.images.top |
视频 (video) | 480×640×3 | 顶部摄像头图像,AV1编码,30 FPS,无深度图属性 |
observation.images.wrist |
视频 (video) | 480×640×3 | 腕部摄像头图像,AV1编码,30 FPS,无深度图属性 |
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(共1个任务) |
机器人类型
- 机器人型号:so_follower
可视化
- 可通过 LeRobot 可视化工具 直观查看数据集内容。



