ankile/real01c-insert-marker-d1-manifest-eval-r1-freecf-bciql-random-arena-sobol20
收藏资源简介:
该数据集是一个机器人数据集,使用LeRobot工具创建,专门用于机器人学习和控制任务。数据集基于Franka机器人,包含60个episodes,总计15662帧数据,覆盖1个任务。数据以parquet格式存储,总数据文件大小为100MB,视频文件大小为200MB,视频帧率为15fps。数据集特征丰富,包括观测状态(如笛卡尔位置、速度、关节位置、速度、夹爪位置)、动作(如笛卡尔速度、位置、关节速度、位置、夹爪动作)、以及元数据如步骤剩余、干预标志、成功标志、奖励、完成标志等。此外,数据集包含来自两个摄像头的图像数据(分辨率480x640,RGB格式),用于视觉观测。数据适用于机器人强化学习、模仿学习等研究。
This is a robotic dataset developed using the LeRobot toolkit, specifically tailored for robot learning and control tasks. Based on the Franka robotic arm, the dataset contains 60 episodes with a total of 15662 frames of data, covering a single task. The core data is stored in Parquet format with a total size of 100 MB, while the accompanying video files have a total size of 200 MB at a frame rate of 15 fps. The dataset provides rich features, including observation states (e.g., Cartesian positions, velocities, joint positions, velocities, and gripper positions), actions (e.g., Cartesian velocities, positions, joint velocities, positions, and gripper actions), as well as metadata such as remaining steps, intervention flags, success flags, rewards, and completion flags. Additionally, the dataset includes image data from two cameras, with a resolution of 480×640 in RGB format for visual observation purposes. This dataset is well-suited for research in areas such as robotic reinforcement learning and imitation learning.




