robot-learning-team43/so101_HQ_merged_diverse_pos

这是一个机器人学习数据集，使用LeRobot创建。数据集包含233个episodes，总计255,511帧，帧率为30fps。特征包括6维动作向量（对应肩部、肘部、腕部和抓手的关节位置）、6维观测状态向量（与动作向量相同）、前端摄像头图像（分辨率为480x640，3通道RGB视频，编码为av1，无音频）、时间戳、帧索引、episode索引、索引和任务索引。数据以parquet文件格式存储，视频文件以mp4格式存储，总数据大小约为100MB，视频大小约为200MB。数据集适用于机器人控制、模拟或强化学习任务，许可证为Apache-2.0。

This is a robotics learning dataset created using LeRobot. It contains 233 episodes with a total of 255,511 frames at 30fps. Features include a 6-dimensional action vector (corresponding to joint positions for shoulder, elbow, wrist, and gripper), a 6-dimensional observation state vector (same as the action vector), front camera images (resolution 480x640, 3-channel RGB video, encoded as av1, no audio), timestamps, frame index, episode index, index, and task index. Data is stored in parquet format, with video files in mp4 format. The total data size is approximately 100MB, and video size is approximately 200MB. The dataset is suitable for robotics control, simulation, or reinforcement learning tasks, licensed under Apache-2.0.