eval_ID_Box_6
收藏资源简介:
该数据集是一个机器人学领域的演示数据集,使用LeRobot框架创建,旨在为机器人学习任务提供真实世界的交互数据。数据内容包含机器人执行任务时的多模态观测记录和相应的控制指令。具体规模为9个任务片段(episodes),共10935帧数据,涵盖1种任务类型,数据以分块形式组织,数据文件总大小约100MB,视频文件总大小约200MB,采集帧率为30fps。结构化特征包括:动作(action)为7维浮点向量,描述机器人的线速度、角速度和夹爪位置;状态观测(observation.state)为7维浮点向量,描述机器人的位置和四元数姿态;还提供来自三个相机(camera1, camera2, camera3)的图像观测,分辨率分别为720x720和720x1280,均为彩色(RGB)视频流。数据集还包含时间戳、帧索引、片段索引、全局索引和任务索引等元数据字段,便于数据对齐和序列分析。该数据集适用于机器人模仿学习、强化学习、视觉伺服控制等研究任务。
This is a robotics demonstration dataset developed using the LeRobot framework, aiming to provide real-world interaction data for robot learning tasks. The dataset includes multimodal observation records and corresponding control commands collected when the robot performs tasks. Specifically, it contains 9 task episodes, totaling 10935 frames, covering 1 type of task. The data is organized in chunks, with a total size of approximately 100MB for data files and 200MB for video files, and the acquisition frame rate is 30fps. Structured features include: the action is a 7-dimensional floating-point vector describing the robot's linear velocity, angular velocity, and gripper position; the state observation (observation.state) is a 7-dimensional floating-point vector describing the robot's position and quaternion attitude; image observations from three cameras (camera1, camera2, camera3) are also provided, with resolutions of 720x720 and 720x1280 respectively, both being RGB video streams. The dataset also includes metadata fields such as timestamps, frame indices, episode indices, global indices, and task indices, which facilitate data alignment and sequence analysis. This dataset is suitable for research tasks such as robot imitation learning, reinforcement learning, and visual servo control.




