salmonHan/record-test_20260528_002139
收藏资源简介:
该数据集是一个机器人操作数据集,使用LeRobot工具创建。数据包含一个总时长为1797帧的单个任务,采样频率为30帧/秒。机器人类型为so_follower,数据以Parquet文件格式存储,总数据文件大小为100MB,视频文件大小为200MB。数据集包含多个特征:动作(action)和观测状态(observation.state)均为6维浮点数组,分别对应机器人关节位置(如肩部平移、肩部提升、肘部弯曲、腕部弯曲、腕部滚动和夹爪位置);观测图像(observation.images.front)为视频格式,分辨率为720x1280,包含3个颜色通道;其他特征包括时间戳、帧索引、任务索引、episode索引和索引。数据分割为训练集(全部数据)。
This is a robotic manipulation dataset constructed using the LeRobot toolkit. It contains a single task with 1797 total frames, sampled at 30 frames per second (fps). The robot platform is so_follower. The structured data is stored in Parquet format, with a total size of 100 MB, while the accompanying video files have a combined size of 200 MB. The dataset includes multiple feature types: both the action and observation.state are 6-dimensional floating-point arrays, which respectively correspond to the robot's joint positions including shoulder translation, shoulder elevation, elbow flexion, wrist flexion, wrist roll, and gripper position; the front observation image (observation.images.front) is stored in video format, with a resolution of 720×1280 and 3 color channels. Additional features include timestamp, frame index, task index, episode index, and index. The full dataset is assigned exclusively to the training split with no reserved validation or test subsets.




