zijian2022/eval_t04
收藏资源简介:
该数据集使用LeRobot创建,专为机器人任务设计,具体针对xarm机器人。数据集包含2个episodes,总计339帧,帧率为20 fps。数据以parquet文件格式存储,总大小为100 MB,视频文件总大小为200 MB。特征包括时间戳、帧索引、episode索引、任务索引、观测状态(7个关节状态,命名为j0到j7)、观测图像(前视和侧视视频,分辨率480x640,3通道,视频编码为av1,帧率30 fps)和动作(7个关节动作,同样命名为j0到j7)。数据集结构支持训练分割(train: 0:2),用于机器人控制和视觉任务的研究。
This dataset was created using LeRobot and is designed for robotics tasks, specifically for the xarm robot. It contains 2 episodes with a total of 339 frames at a frame rate of 20 fps. The data is stored in parquet format with a total size of 100 MB, and video files have a total size of 200 MB. Features include timestamp, frame index, episode index, task index, observation state (7 joint states named j0 to j7), observation images (front and side videos with resolution 480x640, 3 channels, video codec av1, and frame rate 30 fps), and action (7 joint actions also named j0 to j7). The dataset structure supports a training split (train: 0:2) and is intended for research in robot control and vision tasks.




