chris241094/unitree_test_v2

该数据集是使用LeRobot创建的机器人数据集，专门针对Unitree_G1_Brainco机器人。它包含197个episodes，总计220,788帧，覆盖1个任务，并包括788个视频。数据以30fps的帧率采集，全部用于训练。数据集特征包括：观察状态（26维浮点数组，表示机器人的左右肩、肘、腕和手部关节状态）、动作（同样26维浮点数组，对应关节控制）、四个摄像头图像（左高、右高、左腕、右腕，每个为480x640分辨率、3通道的视频，使用AV1编解码器），以及时间戳、帧索引、episode索引等元数据。数据集结构以parquet文件组织，适用于机器人学习和控制任务。This dataset is a robotics dataset created using LeRobot, specifically for the Unitree_G1_Brainco robot. It contains 197 episodes, totaling 220,788 frames, covering 1 task, and includes 788 videos. Data is collected at 30fps and is entirely used for training. The dataset features include: observation state (a 26-dimensional float32 array representing robot joint states for left and right shoulders, elbows, wrists, and hands), action (a similarly 26-dimensional float32 array for joint control), four camera images (left high, right high, left wrist, right wrist, each as 480x640 resolution, 3-channel videos with AV1 codec), and metadata such as timestamp, frame index, and episode index. The dataset is organized in parquet files and is suitable for robotics learning and control tasks.