jae0311/serving_20260527_140234
收藏资源简介:
该数据集是一个机器人控制数据集,使用LeRobot创建,包含267个episodes和153,662帧数据。数据集结构包括动作(action)、观测状态(observation.state)、顶部和手腕摄像头图像(observation.images.top和wrist)、时间戳、帧索引、episode索引等特征。动作和观测状态涉及6个自由度机械臂的关节位置(如肩部平移、肩部提升、肘部弯曲、手腕弯曲、手腕旋转和夹爪位置)。图像数据为视频格式,分辨率480x640,每秒30帧,使用AV1编码。机器人类型为omx_follower,数据以parquet和mp4格式存储,总数据文件大小为100MB,视频文件大小为200MB。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 robotics control dataset created using LeRobot, containing 267 episodes and 153,662 frames. The dataset structure includes features such as action, observation.state, observation.images.top and wrist, timestamp, frame index, episode index, and others. Actions and observation states involve 6-degree-of-freedom robotic arm joint positions (e.g., shoulder pan, shoulder lift, elbow flex, wrist flex, wrist roll, and gripper position). Image data is in video format with a resolution of 480x640, 30 frames per second, using AV1 encoding. The robot type is omx_follower, and data is stored in parquet and mp4 formats, with total data file size of 100MB and video file size of 200MB. The dataset is suitable for robotics learning tasks, such as imitation learning or reinforcement learning.




