so101_placeinbin_blue_20260503_153559_clean
收藏Hugging Face2026-05-19 更新2026-05-19 收录
下载链接:
https://huggingface.co/datasets/RobotLearningProject/so101_placeinbin_blue_20260503_153559_clean
下载链接
链接失效反馈官方服务:
资源简介:
该数据集是一个机器人学数据集,专门设计用于机器人学习和控制任务,特别是与放置物品到垃圾桶(placeinbin)相关的操作。数据集由LeRobot工具创建,包含30个完整的任务片段,总计22076帧数据,帧率为30fps。数据以Parquet格式存储,并附带视频文件。数据集的结构包括多个特征:动作特征(action)表示机器人的6个关节位置(如肩部平移、肩部提升、肘部弯曲、腕部弯曲、腕部旋转和夹爪位置);观察状态特征(observation.state)同样包含这6个关节位置;前视图像特征(observation.images.front)提供720x1280分辨率的RGB视频数据,用于视觉输入。此外,还包括时间戳、帧索引、片段索引、任务索引等元数据。数据集适用于训练和评估机器人控制算法,支持强化学习或模仿学习任务,重点关注机器人在真实或模拟环境中的操作能力。数据集的许可证为Apache-2.0,但主页和论文信息暂未提供。
This dataset is a robotics dataset specifically designed for robot learning and control tasks, particularly related to placing items into a bin (placeinbin). It is created using the LeRobot tool and contains 30 complete task segments, totaling 22,076 frames of data at a frame rate of 30fps. The data is stored in Parquet format and comes with video files. The dataset structure includes multiple features: action features representing the robots 6 joint positions (such as shoulder translation, shoulder lift, elbow bend, wrist bend, wrist rotation, and gripper position); observation state features also include these 6 joint positions; front-view image features provide RGB video data at 720x1280 resolution for visual input. Additionally, metadata such as timestamps, frame indices, segment indices, and task indices are included. The dataset is suitable for training and evaluating robot control algorithms, supporting tasks like reinforcement learning or imitation learning, with a focus on the robots operational capabilities in real or simulated environments. The datasets license is Apache-2.0, but homepage and paper information are not provided.
提供机构:
RobotLearningProject
创建时间:
2026-05-19
原始信息汇总
数据集概述:so101_placeinbin_blue_20260503_153559_clean
- 数据集地址:https://huggingface.co/datasets/RobotLearningProject/so101_placeinbin_blue_20260503_153559_clean
- 许可证:Apache-2.0
- 任务类别:机器人学(robotics)
- 标签:LeRobot
数据集构成
- 总片段数:30
- 总帧数:22076
- 总任务数:1
- 帧率:30 FPS
- 数据文件大小:100 MB
- 视频文件大小:200 MB
- 分块大小:1000
- 数据分割:训练集包含全部30个片段(
train: 0:30)
数据特征
- 动作(action):6维浮点数组,包含机械臂关节位置(shoulder_pan.pos, shoulder_lift.pos, elbow_flex.pos, wrist_flex.pos, wrist_roll.pos, gripper.pos)
- 观察状态(observation.state):6维浮点数组,与动作结构相同
- 观察图像(observation.images.front):视频数据,分辨率720x1280,3通道,编码为AV1,帧率30 FPS
- 时间戳(timestamp):浮点数,形状 [1]
- 帧索引(frame_index):整数,形状 [1]
- 片段索引(episode_index):整数,形状 [1]
- 索引(index):整数,形状 [1]
- 任务索引(task_index):整数,形状 [1]
数据格式
- 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 - 机器人类型:so_follower
其他信息
- 首页:暂无(待补充)
- 论文:暂无(待补充)
- 引用:暂无(待补充)



