so101_grab_red_cube

该数据集将来自`Tank-123/smolvla_test_0129`的数值时间序列转换为Apache TsFile格式，模态为时间序列。原始Hugging Face数据集还包括视频流，但视频未复制到此TsFile存储库中。源数据集使用LeRobot创建，包含SO-101 follower机器人执行任务Grab the red cube的演示。具体信息包括：原始数据集为`Tank-123/smolvla_test_0129`，机器人类型为`so101_follower`，任务为Grab the red cube，episodes数量为72，帧数/TsFile行数为26,610，采样率为30 fps，任务数为1，源分割为`train`（`0:72`），许可证为apache-2.0。转换后的数据存储在一个TsFile中，包含扁平化的向量特征（如action和observation.state），并保留了元数据。视频未包含在此存储库中，但可通过episode_index、frame_index和Time与原始视频帧对齐。

This dataset converts numerical time series from `Tank-123/smolvla_test_0129` into Apache TsFile format, with the modality being time series. The original Hugging Face dataset also includes video streams; videos are not copied into this TsFile repository. The source dataset is created using LeRobot and contains demonstrations of the SO-101 follower robot performing the task Grab the red cube. Specific details: the original dataset is `Tank-123/smolvla_test_0129`, robot type is `so101_follower`, task is Grab the red cube, number of episodes is 72, number of frames/TsFile rows is 26,610, sampling rate is 30 fps, number of tasks is 1, source split is `train` (`0:72`), license is apache-2.0. The converted data is stored in a TsFile containing flattened vector features (such as action and observation.state), with metadata preserved. Videos are not included in this repository but can be aligned with original video frames via episode_index, frame_index, and Time.