kmg0620/pick_battery_bomb_v3_merged

该数据集是一个机器人操作数据集，使用LeRobot创建，专门用于机器人技能学习。数据集包含204个episodes，总计162,078帧，涉及2个任务。数据以30fps的帧率采集，存储为parquet格式，总数据量约为2000MB，视频文件量极小。机器人类型为bi_so_follower（可能表示双手机器人跟随器）。特征包括动作（action）和观察（observation），其中动作包含12个浮点数，代表左右机械臂的关节位置（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）；观察包括状态（同样12个关节位置）和图像，图像来自两个摄像头视角：left_main（左侧主摄像头）和right_right_wrist（右侧腕部摄像头），分辨率均为240x320，3通道RGB视频。其他特征包括时间戳、帧索引、episode索引、索引和任务索引。数据集仅包含训练分割，适用于机器人控制、模仿学习或强化学习任务。

This dataset is a robotic manipulation dataset created using LeRobot, specifically designed for robot skill learning. It contains 204 episodes, totaling 162,078 frames, and involves 2 tasks. Data is collected at 30fps, stored in parquet format, with a total data size of approximately 2000MB and minimal video file size. The robot type is bi_so_follower (likely indicating a bimanual robot follower). Features include actions and observations: actions consist of 12 float values representing joint positions for both left and right arms (e.g., shoulder pan, shoulder lift, elbow flex, wrist flex, wrist roll, and gripper positions); observations include state (same 12 joint positions) and images from two camera perspectives: left_main (left main camera) and right_right_wrist (right wrist camera), both with 240x320 resolution and 3-channel RGB video. Additional features include timestamp, frame index, episode index, index, and task index. The dataset only includes a training split and is suitable for robotic control, imitation learning, or reinforcement learning tasks.