kunhsiang/eval_exp9_sc1var_grasp_the_small_box_20260528-183227
收藏资源简介:
该数据集由LeRobot创建,专注于机器人技术任务,特别是针对一种名为so_follower的机器人类型。数据集包含一个完整的训练情节,总计2563帧,数据以parquet文件格式存储,视频文件以mp4格式存储。数据集特征包括机器人的动作(如肩部、肘部、腕部和夹爪的位置)、观察状态(包括来自前部、顶部和夹爪视角的图像视频,分辨率为480x640,30fps),以及时间戳、帧索引、情节索引等元数据。数据集设计用于机器人控制任务,例如抓取小盒子的场景,适用于训练和评估机器人学习模型。
This dataset was created using LeRobot and focuses on robotics tasks, specifically for a robot type called so_follower. It contains one full training episode with a total of 2563 frames, stored in parquet files for data and mp4 files for videos. The dataset features include robot actions (e.g., positions of shoulder, elbow, wrist, and gripper), observation states (including video images from front, top, and gripper perspectives at 480x640 resolution and 30fps), and metadata such as timestamp, frame index, and episode index. It is designed for robotics control tasks, such as grasping a small box, and is suitable for training and evaluating robot learning models.




