xiaochyVera/long_horizon_pick_toys_test

该数据集是一个机器人控制数据集，使用LeRobot工具创建。数据集包含3个episodes，总计1354帧，涉及1个任务，并包含9个视频。数据以parquet格式存储，帧率为30 fps。数据集的特征包括：动作（8维浮点数，表示7个关节和夹爪控制）、观测状态（8维浮点数，表示关节状态）、右末端执行器姿态（10维浮点数，表示6D旋转、3D平移和夹爪关节）、来自Azure Kinect左摄像头的深度图像（720x1280x1）和彩色图像（720x1280x3）、来自Azure Kinect前摄像头的彩色图像（720x1280x3），以及时间戳、帧索引、episode索引等元数据。数据集适用于机器人学习任务，如模仿学习或强化学习。

This dataset is a robotics control dataset created using the LeRobot tool. It contains 3 episodes, totaling 1354 frames, involving 1 task, and includes 9 videos. The data is stored in parquet format with a frame rate of 30 fps. The dataset features include: action (8-dimensional float32, representing 7 joints and gripper control), observation state (8-dimensional float32, representing joint states), right end-effector pose (10-dimensional float32, representing 6D rotation, 3D translation, and gripper articulation), depth image (720x1280x1) and color image (720x1280x3) from the Azure Kinect left camera, color image (720x1280x3) from the Azure Kinect front camera, as well as metadata such as timestamp, frame index, and episode index. The dataset is suitable for robotics learning tasks, such as imitation learning or reinforcement learning.