so101_duct_tape_pick_place
收藏资源简介:
该数据集是使用LeRobot框架创建的机器人数据集,适用于机器人相关任务。数据集包含25个episodes,总计10729帧,对应一个任务。数据以Parquet文件格式存储,总数据文件大小约为100 MB,关联的视频文件大小约为200 MB,视频帧率为30 FPS。数据集仅包含训练集划分。核心特征包括:1) 动作(action):一个6维浮点向量,代表机器人关节位置(肩部平移、肩部升降、肘部弯曲、腕部弯曲、腕部旋转、夹持器位置);2) 观察状态(observation.state):一个与动作结构相同的6维浮点向量,表示机器人当前关节状态;3) 图像观察:来自前置摄像头(observation.images.front)和腕部摄像头(observation.images.wrist)的视频数据,每个视频帧的分辨率为640x480,3通道彩色,使用h264编解码器;4) 元数据字段:包括时间戳、帧索引、episode索引、数据索引和任务索引,用于标识数据点。数据集采用Apache 2.0许可证。
This dataset is created using the LeRobot framework for robot-related tasks. It contains 25 episodes, totaling 10729 frames, corresponding to one task. The data is stored in Parquet format, with a total data file size of approximately 100 MB and associated video files of about 200 MB, with a video frame rate of 30 FPS. The dataset includes only a training set split. Key features include: 1) Action: a 6-dimensional floating-point vector representing robot joint positions (shoulder translation, shoulder lift, elbow bend, wrist bend, wrist rotation, gripper position); 2) Observation state (observation.state): a 6-dimensional floating-point vector with the same structure as action, indicating the current robot joint state; 3) Image observations: video data from a front camera (observation.images.front) and a wrist camera (observation.images.wrist), with each video frame having a resolution of 640x480, 3-channel color, using the h264 codec; 4) Metadata fields: including timestamp, frame index, episode index, data index, and task index for identifying data points. The dataset is licensed under Apache 2.0.




