stevenworkspace/eval_take_mvactd_13
收藏资源简介:
该数据集使用LeRobot创建,专注于机器人技术领域,具体针对移动AI机器人。数据集包含1个完整episode、2893帧和1个任务,数据以parquet文件格式组织,总数据文件大小为100MB,视频文件大小为200MB,帧率为30fps。数据集的特征包括动作(action)和观察(observation),其中动作由16个浮点数值组成,表示左右机械臂的关节位置和速度;观察包括状态(同样为16个浮点数值,与动作对应)和来自三个摄像头的图像视频数据:高位摄像头(cam_high)、左腕摄像头(cam_left_wrist)和右腕摄像头(cam_right_wrist),所有视频分辨率为480x640,编码为av1,无音频。此外,数据集还包含时间戳、帧索引、episode索引等元数据。该数据集适用于机器人控制、视觉感知和强化学习等任务。
This dataset was created using LeRobot and focuses on the robotics domain, specifically for a mobile AI robot. It contains 1 full episode, 2893 frames, and 1 task, organized in parquet format with a total data file size of 100MB and video file size of 200MB, at a frame rate of 30fps. The dataset features include action and observation, where action consists of 16 float values representing joint positions and velocities for the left and right robotic arms; observation includes state (also 16 float values, corresponding to action) and image video data from three cameras: a high camera (cam_high), left wrist camera (cam_left_wrist), and right wrist camera (cam_right_wrist), all with a resolution of 480x640, encoded in av1, and without audio. Additionally, the dataset includes metadata such as timestamp, frame index, episode index, etc. It is suitable for tasks like robot control, visual perception, and reinforcement learning.




