leokswang/22122025-foldclo-06_lerobot_format
收藏资源简介:
该数据集是使用LeRobot工具创建的机器人学习数据集,专门用于Yam机器人。数据集包含10个训练集,总计16817帧,帧率为30fps。数据以Parquet文件格式存储,每个文件包含多视角图像观测(顶部、左侧、右侧摄像头,分辨率为360x640,3通道)、机器人状态观测(14维浮点数组)、动作(14维浮点数组)以及时间戳、帧索引、集索引、索引和任务索引等元数据。数据集没有视频文件,仅包含结构化数据,适用于机器人控制、强化学习或模仿学习任务。数据分割为训练集,覆盖所有10个集。
This dataset is a robotics learning dataset created using the LeRobot tool, specifically for the Yam robot. It contains 10 training episodes with a total of 16817 frames at 30fps. The data is stored in Parquet format, each file including multi-view image observations (top, left, right camera views at 360x640 resolution, 3 channels), robot state observations (14-dimensional float32 array), actions (14-dimensional float32 array), and metadata such as timestamp, frame index, episode index, index, and task index. The dataset does not include video files, only structured data, making it suitable for robotics control, reinforcement learning, or imitation learning tasks. The data is split into a training set covering all 10 episodes.




