nirs123/pick_blue_cube_90_episodes
收藏资源简介:
该数据集是一个机器人操作数据集,专注于拾取蓝色立方体任务,使用LeRobot工具创建。数据集包含90个训练episodes,总计51230帧,帧率为30fps。数据特征包括:动作空间(6维浮点数组,表示肩部平移、肩部提升、肘部弯曲、手腕弯曲、手腕旋转和夹爪位置)、状态观测(与动作相同的6维关节位置)、来自手腕摄像头(分辨率640x480)和头顶摄像头(分辨率480x640)的视频观测(均为RGB格式,30fps),以及时间戳、帧索引、episode索引等元数据。机器人类型为so_follower。数据以parquet文件(总大小约100MB)和mp4视频文件(总大小约200MB)格式存储,分割为训练集(所有90个episodes)。数据集适用于机器人学习任务,如模仿学习或强化学习。
This dataset is a robotic manipulation dataset focused on the pick blue cube task, created using the LeRobot tool. It contains 90 training episodes with a total of 51,230 frames at 30 fps. Features include: action space (6-dimensional float array representing shoulder pan, shoulder lift, elbow flex, wrist flex, wrist roll, and gripper positions), state observation (same 6-dimensional joint positions as actions), video observations from a wrist camera (640x480 resolution) and an overhead camera (480x640 resolution), both in RGB format at 30 fps, along with metadata such as timestamp, frame index, and episode index. The robot type is so_follower. Data is stored in parquet files (total size ~100MB) and mp4 video files (total size ~200MB), split into a training set (all 90 episodes). The dataset is suitable for robotic learning tasks, such as imitation learning or reinforcement learning.




