leokswang/22122025-foldclo-02_lerobot_format
收藏资源简介:
该数据集是一个用于机器人任务的数据集,基于LeRobot框架创建,专门针对YAM机器人。数据集包含10个episodes,总计21629帧,帧率为30fps。数据以分块形式存储,每个块大小为1000。数据集没有视频文件,但提供了多视角的图像观测,包括顶部、左侧和右侧摄像头图像,每个图像的分辨率为360x640x3。此外,数据集包含观测状态(14维浮点数组)、动作(14维浮点数组)、时间戳、帧索引、episode索引、索引和任务索引等特征。这些数据适用于机器人控制、强化学习或行为克隆等任务。数据集采用Apache 2.0许可证,但主页和论文信息暂未提供。
This is a robotics task dataset developed on the LeRobot framework, exclusively designed for the YAM robot. It consists of 10 episodes, with a total of 21,629 frames at a frame rate of 30 fps. The dataset is stored in chunks, each with a size of 1000. No video files are included in the dataset; instead, multi-view image observations are provided, covering top, left, and right camera images, each with a resolution of 360×640×3. Additionally, the dataset contains various features including observation states (14-dimensional floating-point arrays), actions (14-dimensional floating-point arrays), timestamps, frame indices, episode indices, indices, and task indices. This dataset is suitable for applications such as robot control, reinforcement learning, and behavioral cloning. The dataset is released under the Apache 2.0 license, while its official homepage and associated paper information are not yet available.




