robocasa_test1

该数据集是一个用于机器人学习任务的数据集，使用LeRobot工具创建。数据集包含机器人观察图像（如手眼摄像头和左右视角摄像头，分辨率为256x256，3通道，视频格式为h264）、任务描述、任务名称、状态观察（16维浮点数）、奖励（浮点数）、完成标志（布尔值）、时间戳、帧索引、回合索引、索引、任务索引以及动作（12维浮点数）等特征。数据以每秒20帧的速率采集，总共有1个回合、274帧和49个任务。数据文件为Parquet格式，视频文件为MP4格式，使用PandaOmron机器人类型。该数据集适用于机器人控制、强化学习和计算机视觉研究。This is a dataset for robotic learning tasks, created using the LeRobot toolkit. The dataset includes robotic observation images (such as eye-in-hand and left/right perspective cameras, with a resolution of 256×256, 3 channels, and the video format is h264), task descriptions, task names, state observations (16-dimensional floating-point numbers), rewards (floating-point numbers), completion flags (boolean values), timestamps, frame indices, episode indices, overall indices, task indices, and actions (12-dimensional floating-point numbers). The data is collected at a rate of 20 frames per second, with a total of 1 episode, 274 frames, and 49 tasks. The data files are in Parquet format, while the video files are in MP4 format. This dataset adopts the PandaOmron robot type and is suitable for research in robotic control, reinforcement learning, and computer vision.