bartek-niedzielski/fast_pick_and_place_40

这是一个机器人学数据集，使用LeRobot创建，专为机器人控制和研究设计。数据集包含40个episodes，总计10121帧和80个视频，以15fps的帧率录制。数据以parquet格式存储，并分为一个训练集（splits: train: 0:40）。特征包括外部图像（exterior_image_1_left）和手腕图像（wrist_image_left），均为480x640x3的视频格式，使用av1编解码器；关节位置（joint_position）为7维浮点数组，表示机器人的7个关节角度；夹爪位置（gripper_position）为1维浮点数组；动作（actions）和状态（state）均为8维浮点数组，结合了关节和夹爪信息；此外还有时间戳（timestamp）、帧索引（frame_index）、episode索引（episode_index）、索引（index）和任务索引（task_index）。数据集适用于机器人学习、控制算法开发和模拟任务。This is a robotics dataset created using LeRobot, designed for robot control and research. It contains 40 episodes, totaling 10121 frames and 80 videos, recorded at 15fps. The data is stored in parquet format and includes one training split (splits: train: 0:40). Features include exterior images (exterior_image_1_left) and wrist images (wrist_image_left), both in video format with dimensions 480x640x3 and using the av1 codec; joint positions (joint_position) as a 7-dimensional float array representing robot joint angles; gripper position (gripper_position) as a 1-dimensional float array; actions (actions) and state (state) as 8-dimensional float arrays combining joint and gripper information; along with timestamp (timestamp), frame index (frame_index), episode index (episode_index), index (index), and task index (task_index). The dataset is suitable for robot learning, control algorithm development, and simulation tasks.