dot_collect_empty_bottle_black_white_wrist_200k_bs8_k48_lba0_testing0

这是一个使用LeRobot创建的机器人学数据集，专门用于机器人控制和学习任务。数据集包含1个完整的情节（episode），总计1309帧数据，对应1个特定任务和1个视频记录。数据以parquet格式存储，包含丰富的特征信息：9维动作向量（机械臂各关节位置和末端执行器速度）、9维观测状态（与动作相同的机械臂状态）、480x640分辨率的手腕摄像头RGB图像、时间戳、帧索引、情节索引等元数据。数据集采用30fps的采样频率，适用于机器人控制策略学习和模仿学习研究。

This is a robotics dataset constructed using LeRobot, specifically designed for robotic control and learning tasks. The dataset contains 1 complete episode with a total of 1309 frames of data, corresponding to one specific task and one video recording. The data is stored in Parquet format and includes rich feature information: a 9-dimensional action vector (robotic arm joint positions and end-effector velocity), a 9-dimensional observation state (the same robotic arm state as the action), 480x640 resolution RGB images from the wrist camera, timestamps, frame indices, episode indices and other metadata. The dataset uses a 30fps sampling frequency and is suitable for research on robotic control policy learning and imitation learning.