zhang311/testlerobot2
收藏资源简介:
该数据集是一个机器人数据集,使用LeRobot创建,专为机器人任务设计。数据集包含100个episodes,总计57,661帧,帧率为30fps,采用机器人类型bi_so101_follower。数据集结构包括动作特征(12维浮点数,表示左右机械臂的关节位置,如肩部平移、肩部提升、肘部弯曲、腕部弯曲、腕部旋转和夹爪位置)、观测状态特征(12维浮点数,与动作特征相同),以及三个摄像头图像观测(左摄像头:480x640分辨率,顶部摄像头:360x640分辨率,右摄像头:480x640分辨率,均为彩色视频,使用h264编码)。此外,数据集还包括时间戳、帧索引、episode索引、总索引和任务索引等元数据。数据以parquet文件格式存储,视频文件为mp4格式,总数据文件大小为100MB,视频文件大小为200MB。数据集仅包含训练分割(0到100个episodes),适用于机器人控制、模仿学习或强化学习任务。
This dataset is a robotics dataset created using LeRobot, designed for robotics tasks. It contains 100 episodes with a total of 57,661 frames at 30 fps, using the robot type bi_so101_follower. The dataset structure includes action features (12-dimensional float32 values representing joint positions for left and right arms, such as shoulder pan, shoulder lift, elbow flex, wrist flex, wrist roll, and gripper positions), observation state features (12-dimensional float32 values, same as action features), and three camera image observations (left camera: 480x640 resolution, top camera: 360x640 resolution, right camera: 480x640 resolution, all in color video with h264 encoding). Additionally, the dataset includes metadata such as timestamp, frame index, episode index, overall index, and task index. Data is stored in parquet file format, with video files in mp4 format; the total data file size is 100MB, and video file size is 200MB. The dataset only includes a training split (episodes 0 to 100) and is suitable for robotics control, imitation learning, or reinforcement learning tasks.




