eval_pov_mixed
收藏资源简介:
该数据集是使用LeRobot工具创建的机器人学数据集,专门针对so_follower类型的机器人。它以结构化格式记录了机器人的交互数据,包括:机器人的动作指令(涵盖肩部平移、肩部抬升、肘部弯曲、腕部弯曲、腕部旋转和夹爪位置共6个关节的控制量);观测状态(与动作对应的6个关节位置反馈);来自前摄像头的图像观测(视频格式,分辨率为480x640像素,3通道彩色,帧率为30fps);以及时间戳、帧索引、回合索引和任务索引等元数据。数据以Parquet文件格式组织存储,并按照分块进行管理(每块约1000个数据点)。数据总规模在示例中未具体说明,但提供了数据文件约100MB、视频文件约200MB的参考信息。该数据集适用于机器人控制、模仿学习、视觉伺服等机器人学相关任务的研究与开发。
This dataset is a robotics dataset created using the LeRobot tool, specifically designed for the so_follower type of robot. It records robot interaction data in a structured format, including: robot action commands (covering control inputs for 6 joints: shoulder translation, shoulder elevation, elbow flexion, wrist flexion, wrist rotation, and gripper position); observation states (corresponding 6-joint position feedback); image observations from a front camera (video format, resolution 480x640 pixels, 3-channel color, frame rate 30fps); and metadata such as timestamps, frame indices, episode indices, and task indices. The data is organized and stored in Parquet file format and managed in chunks (approximately 1000 data points per chunk). The total data size is not specified in the example, but reference information is provided indicating data files of about 100MB and video files of about 200MB. This dataset is suitable for research and development in robotics-related tasks such as robot control, imitation learning, and visual servoing.




