simple-pick-and-place
收藏Hugging Face2026-05-20 更新2026-05-20 收录
下载链接:
https://huggingface.co/datasets/Exxxx03/simple-pick-and-place
下载链接
链接失效反馈官方服务:
资源简介:
该数据集是使用LeRobot工具创建的,专注于机器人学领域,具体针对‘so_follower’机器人类型。数据集包含31个episodes,总计10041帧,数据以Parquet格式存储,并配有视频文件(总大小约300MB)。数据特征包括机器人的动作(如关节位置)、观察状态(关节位置)、来自前方和手腕摄像头的图像观察(分辨率为480x640,3通道,30fps),以及补充信息如策略动作、干预标志、状态、收集器策略ID、时间戳、帧索引等。数据集结构详细,适用于机器人控制、模仿学习或强化学习任务,但主页、论文和引用信息暂缺。
This dataset is created using the LeRobot tool, focusing on the field of robotics, specifically for the so_follower robot type. It contains 31 episodes, totaling 10,041 frames, stored in Parquet format along with video files (total size approximately 300MB). Data features include robot actions (e.g., joint positions), observation states (joint positions), image observations from front and wrist cameras (resolution 480x640, 3 channels, 30fps), and supplementary information such as policy actions, intervention flags, states, collector policy IDs, timestamps, and frame indices. The dataset is structured in detail, suitable for robot control, imitation learning, or reinforcement learning tasks, but homepage, paper, and citation information are currently missing.
提供机构:
Exxxx03
创建时间:
2026-05-20
原始信息汇总
数据集概述
- 数据集名称: simple-pick-and-place
- 许可证: Apache-2.0
- 任务类别: 机器人学(Robotics)
- 创建工具: 使用 LeRobot 创建
- 机器人类型: so_follower
数据集规模
- 总片段数(Episodes): 31
- 总帧数(Frames): 10041
- 总任务数(Tasks): 1
- 帧率(FPS): 30
- 数据文件大小: 100 MB
- 视频文件大小: 200 MB
- 数据分块大小: 1000 帧/块
- 数据集划分:
- 训练集(train): 片段 0 至 30(共 31 个片段)
数据格式
- 数据文件路径:
data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet - 视频文件路径:
videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4 - 视频编码: AV1
- 视频像素格式: yuv420p
- 视频分辨率: 480×640(高度×宽度)
- 视频通道数: 3
特征结构
该数据集包含以下特征字段:
| 特征名称 | 数据类型 | 形状 | 描述 |
|---|---|---|---|
action |
float32 | (6,) | 机器人动作指令,包含肩关节、肘关节、腕关节及夹爪位置 |
observation.state |
float32 | (6,) | 机器人观测状态,与动作指令维度一致 |
observation.images.front |
视频 | (480, 640, 3) | 前置摄像头图像 |
observation.images.wrist |
视频 | (480, 640, 3) | 腕部摄像头图像 |
complementary_info.policy_action |
float32 | (6,) | 策略输出的动作 |
complementary_info.is_intervention |
float32 | (1,) | 是否为人工干预标识 |
complementary_info.state |
float32 | (1,) | 补充状态信息 |
complementary_info.collector_policy_id |
string | (1,) | 数据采集策略ID |
complementary_info.value_r0 |
float32 | (1,) | 值函数(r0) |
complementary_info.advantage_r0 |
float32 | (1,) | 优势函数(r0) |
complementary_info.acp_indicator_r0 |
int64 | (1,) | ACP指示器(r0) |
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 |
动作/状态字段名称(6维依次为):
shoulder_pan.pos(肩部旋转位置)shoulder_lift.pos(肩部抬升位置)elbow_flex.pos(肘部弯曲位置)wrist_flex.pos(腕部弯曲位置)wrist_roll.pos(腕部旋转位置)gripper.pos(夹爪位置)



