JennyWWW/splatsim_approach_lever_11_biasend_5path

该数据集是一个机器人领域的数据集，使用LeRobot创建。数据集包含300个回合，总计81392帧，专注于1个任务。数据以parquet格式存储，视频以mp4格式存储，帧率为30fps。数据集特征包括观察图像（基础RGB和手腕RGB的两种处理方式，均为3通道224x224图像）、观察状态（7维浮点数，表示关节1-6和夹持器）、动作（7维浮点数，与观察状态对应）、时间戳、帧索引、回合索引、索引和任务索引。数据集结构分为训练集（包含所有300个回合），适用于机器人控制和学习任务。This dataset is a robotics dataset created using LeRobot. It contains 300 episodes, totaling 81,392 frames, and focuses on 1 task. The data is stored in parquet format, with videos in mp4 format at 30 fps. The dataset features include observation images (base RGB and wrist RGB in both letterbox and stretch variants, each as 3-channel 224x224 images), observation state (7-dimensional float32 representing joints 1-6 and gripper), action (7-dimensional float32, corresponding to the observation state), timestamp, frame index, episode index, index, and task index. The dataset structure includes a training split covering all 300 episodes, suitable for robot control and learning tasks.