RoboCOIN/AI2_Alphabot_2_move_test_tube
收藏资源简介:
该数据集基于LeRobot格式扩展,并完全兼容LeRobot框架。它是一个机器人操作数据集,专注于用一只手将试管重新定位到试管架的不同槽位的任务。数据集包含69个episodes,总计33010帧,大小为2.93 GB,帧率为30 FPS。使用名为AI2_Alphabot_2的机器人,配备双指末端执行器,并通过VR控制器进行遥操作。传感器包括4个RGB摄像头:前胸摄像头、前头摄像头、左手腕摄像头和右手腕摄像头,分辨率均为640x480,视频编码为AV1。数据集结构包括视频文件、机器人状态数据(如关节位置、末端执行器位置和旋转)、动作数据以及丰富的注释(如末端执行器方向、速度和夹爪活动)。数据组织为parquet格式,分为训练集(episodes 0到68)。该数据集由北京智源人工智能研究院的RoboCOIN团队贡献,适用于机器人学习和模仿学习研究。
This dataset is extended based on the LeRobot format and fully compatible with the LeRobot framework. It is a robotic manipulation dataset focusing on the task of repositioning test tubes into different slots of a test tube rack using a single hand. The dataset contains 69 episodes, totaling 33010 frames, with a size of 2.93 GB and a frame rate of 30 FPS. The robot used is named AI2_Alphabot_2, which is equipped with a two-finger end-effector and teleoperated via a VR controller. The sensors include 4 RGB cameras: the front-chest camera, front-head camera, left-wrist camera and right-wrist camera, all with a resolution of 640×480 and video encoded in AV1. The dataset structure consists of video files, robot state data (such as joint positions, end-effector positions and rotations), action data and rich annotations (such as end-effector orientation, velocity and gripper activity). The data is organized in Parquet format and split into the training set (episodes 0 to 68). This dataset is contributed by the RoboCOIN Team from Beijing Zhiyuan Artificial Intelligence Research Institute, and is applicable to research in robotic learning and imitation learning.




