YOLO2431/put_plate_to_rack_v2_full_keyframe

该数据集是一个使用LeRobot创建的双手机器人（yam_bimanual）演示数据集，专门用于机器人学习任务。数据集包含308个完整的演示序列（episodes），总计318,073帧数据，帧率为每秒30帧。数据集中提供了丰富的多模态信息：包括机器人状态观测（32维浮点向量，涵盖左右机械臂的3D位置、6D旋转表示、夹爪开合状态以及各关节角度）、机器人动作指令（同样为32维浮点向量）、以及来自三个不同视角的视频流（头戴摄像头、左腕摄像头、右腕摄像头），所有视频均为720x1280分辨率、RGB三通道格式，采用H.264编码。数据集还包含时间戳、帧索引、episode索引和任务索引等元数据。数据以parquet文件格式组织，仅包含训练集，适用于机器人模仿学习、强化学习等研究。This dataset is a bimanual robot (yam_bimanual) demonstration dataset created using LeRobot, specifically designed for robotics learning tasks. It contains 308 complete demonstration episodes, totaling 318,073 frames at 30 frames per second. The dataset provides rich multimodal information: including robot state observations (32-dimensional float vectors, covering 3D positions, 6D rotation representations, gripper states, and joint angles for both left and right arms), robot action commands (also 32-dimensional float vectors), and video streams from three different perspectives (head camera, left wrist camera, right wrist camera), all videos are in 720x1280 resolution, RGB three-channel format, using H.264 encoding. The dataset also includes metadata such as timestamps, frame indices, episode indices, and task indices. Data is organized in parquet file format, containing only training splits, suitable for research in robot imitation learning, reinforcement learning, etc.