yoko2606/movethegreenbrick_20260525_203129
收藏资源简介:
这是一个使用LeRobot创建的机器人控制数据集,专门用于机器人学习任务。数据集包含50个训练episodes,总计22447帧,数据来自so_follower机器人类型。它提供了丰富的多模态信息:动作数据包括6个关节位置(如肩部平移、肩部提升、肘部弯曲、腕部弯曲、腕部滚动和夹爪位置),观测数据包括相同的关节状态和顶部摄像头图像(分辨率为480x640,30帧每秒,RGB通道)。数据集以parquet和视频格式存储,支持机器人控制算法的训练和评估,适用于强化学习或模仿学习场景。
This dataset is a robotics control dataset created using LeRobot, designed for robot learning tasks. It contains 50 training episodes with a total of 22447 frames, sourced from a so_follower robot type. The dataset offers rich multimodal information: action data includes 6 joint positions (e.g., shoulder pan, shoulder lift, elbow flex, wrist flex, wrist roll, and gripper positions), while observation data includes the same joint states and top camera images (with a resolution of 480x640, 30 fps, RGB channels). Stored in parquet and video formats, it supports training and evaluation of robot control algorithms, suitable for reinforcement learning or imitation learning scenarios.




