MooreFoss/so101_task_putinto_20260529_151343
收藏资源简介:
该数据集使用LeRobot工具创建,专注于机器人学任务,具体涉及putinto操作。数据集包含30个训练集片段,总计6329帧数据,数据以parquet格式存储,总大小约300MB(数据文件100MB,视频文件200MB)。数据结构包括动作(action)、观测状态(observation.state)、固定摄像头图像(observation.images.fixed)、手眼摄像头图像(observation.images.handeye)等特征,图像分辨率为480x640,帧率为15fps。机器人类型为so_follower,动作和状态特征均包含6个关节位置(如肩部平移、肩部升降、肘部弯曲等)。数据集适用于机器人控制和模仿学习研究。
This dataset is developed using the LeRobot toolkit, focusing on robotics tasks, specifically the "put into" manipulation operation. It includes 30 training episodes, with a total of 6329 frames. The data is stored in Parquet format, with an overall size of approximately 300 MB (100 MB for data files and 200 MB for video files). The dataset structure comprises features such as action, observation.state, fixed camera images (observation.images.fixed), and hand-eye camera images (observation.images.handeye). The image resolution is 480x640, and the frame rate is 15 fps. The robot type used is so_follower. Both the action and state features contain 6 joint positions, such as shoulder translation, shoulder elevation, elbow flexion, and others. This dataset is suitable for research in robot control and imitation learning.




