CarlosMunoz0/vla-mujoco-so101-cube_on_tray-joystick-v1

该数据集是一个机器人控制数据集，使用LeRobot工具创建，专注于模拟环境中的操作任务。数据集基于Mujoco模拟器，机器人类型为so101_mujoco，任务涉及通过操纵杆控制机器人将立方体放在托盘上（cube_on_tray-joystick-v1）。数据集包含60个完整的情节（episodes），总计44995帧，帧率为30fps。数据以parquet格式存储，总数据文件大小为100MB，视频文件大小为200MB。特征包括：动作数据（6维浮点数组，对应机器人关节位置如肩部、肘部、腕部和夹持器）、状态观测（13维浮点数组，包括关节位置和末端执行器的位置与四元数）、以及多个图像观测（如realsense、realsense_depth、realsense_depth_vis和wrist_cam视频，分辨率均为480x640x3）。数据集还包含时间戳、帧索引、情节索引等元数据，适用于机器人学习和控制研究。

This dataset is a robotics control dataset created using the LeRobot tool, focusing on manipulation tasks in a simulated environment. It is based on the Mujoco simulator with a robot type of so101_mujoco, and the task involves controlling a robot via a joystick to place a cube on a tray (cube_on_tray-joystick-v1). The dataset contains 60 complete episodes, totaling 44,995 frames at a frame rate of 30fps. Data is stored in parquet format, with a total data file size of 100MB and video file size of 200MB. Features include: action data (a 6-dimensional float32 array corresponding to robot joint positions such as shoulder, elbow, wrist, and gripper), state observations (a 13-dimensional float32 array including joint positions and end-effector position with quaternion), and multiple image observations (e.g., realsense, realsense_depth, realsense_depth_vis, and wrist_cam videos, all with a resolution of 480x640x3). The dataset also includes metadata such as timestamps, frame indices, and episode indices, making it suitable for robotics learning and control research.