Robo360

Robo360是由加州大学圣地亚哥分校和清华大学合作创建的3D全方位多材料机器人操作数据集。该数据集包含超过2000个多视角机器人操作轨迹，由86个精确校准的相机捕捉，涵盖了100多种具有不同物理和光学特性的物体。Robo360旨在推动3D机器人操作领域的研究，特别是在理解和模拟操作过程中的低级动态。数据集的应用领域包括3D场景表示学习、视觉策略学习和材料建模，旨在提高机器人对物理世界的控制理解。

Robo360 is a 360-degree multimaterial 3D robotic manipulation dataset jointly developed by the University of California, San Diego and Tsinghua University. This dataset includes over 2,000 multi-view robotic manipulation trajectories captured by 86 precisely calibrated cameras, covering more than 100 objects with distinct physical and optical properties. Robo360 is designed to advance research in the field of 3D robotic manipulation, particularly in understanding and simulating low-level dynamics during manipulation processes. The application scenarios of this dataset cover 3D scene representation learning, visual policy learning, and material modeling, with the goal of enhancing robots' understanding and control over the physical world.