shirasu206/dual_arm_ur5e_ur3e_20260530_v6
收藏资源简介:
该数据集是一个用于机器人学的双臂机器人数据集,使用LeRobot创建。数据集包含UR5e和UR3e双臂机器人的观测和动作数据。具体来说,数据集包括10个episodes,总计5838帧,覆盖1个任务。数据以parquet文件格式存储,分割为训练集(episodes 0到10)。观测数据包括三个视频流:base_image、ur5e_image和ur3e_image,每个视频分辨率为224x224像素,3通道(RGB),使用H.264编码,帧率为15fps,无音频。状态观测为14维浮点数向量,分别表示UR5e和UR3e机器人的6个关节位置和夹爪状态。动作为14维浮点数向量,对应两个机器人的关节和夹爪控制。此外,数据集还包含时间戳、帧索引、episode索引、索引和任务索引等元数据。该数据集适用于机器人控制、模仿学习或强化学习任务,许可证为Apache-2.0。
This dataset is a dual-arm robotics dataset created using LeRobot. It contains observation and action data from UR5e and UR3e dual-arm robots. Specifically, the dataset includes 10 episodes, totaling 5838 frames, covering 1 task. The data is stored in parquet format, split into a training set (episodes 0 to 10). Observations consist of three video streams: base_image, ur5e_image, and ur3e_image, each with a resolution of 224x224 pixels, 3 channels (RGB), encoded in H.264, at 15fps, without audio. State observations are 14-dimensional float vectors representing the joint positions and gripper states of both UR5e and UR3e robots. Actions are 14-dimensional float vectors corresponding to joint and gripper controls for both robots. Additionally, the dataset includes metadata such as timestamps, frame indices, episode indices, indices, and task indices. It is suitable for robotics control, imitation learning, or reinforcement learning tasks, licensed under Apache-2.0.




