lbracket-pick-place-v2-300ep_kojima

该数据集是一个专门用于机器人操作任务的L形支架拾放数据集（lbracket-pick-place-v2-300ep_kojima），包含300个episodes，总计121,981帧数据。数据以parquet格式存储，并附带视频文件。数据集特征包括：8维动作向量（控制关节位置和夹爪位置）、24维状态观测向量（涵盖7个关节和夹爪的位置、速度和扭矩信息），以及多视角图像数据（如颈部、右侧腕部、头顶和胸部摄像头的RGB视频和深度图像）。图像分辨率为480x640或480x848，帧率为30fps。数据集使用Apache 2.0许可证，由LeRobot工具创建，适用于机器人学习和强化学习研究。

This dataset is a robot manipulation dataset specifically designed for the L-shaped bracket pick and place task (lbracket-pick-place-v2-300ep_kojima). It contains 300 episodes, totaling 121,981 frames, stored in parquet format with accompanying video files. The dataset features include: an 8-dimensional action vector (controlling joint positions and gripper position), a 24-dimensional state observation vector (covering position, velocity, and torque information for 7 joints and the gripper), and multi-view image data (such as RGB videos and depth images from neck, right wrist, overhead, and chest cameras). The image resolution is 480x640 or 480x848, with a frame rate of 30fps. The dataset is licensed under Apache 2.0, created by the LeRobot tool, and suitable for robotics learning and reinforcement learning research.