Motion Planning Problems Dataset

该数据集包含了在不同环境下随机生成的运动规划问题，总计约327万个问题，跨越了57.5万个环境。此外，该数据集分为三个难度递增的问题类别，并用于训练M$\pi$Nets，以便从深度相机观测中生成平滑的运动。其规模达到327万个问题，分布在超过57.5万个环境中，任务旨在生成无碰撞的运动。

This dataset contains motion planning problems randomly generated across different environments, with a total of approximately 3.27 million problems spanning over 575,000 environments. Additionally, the dataset is divided into three categories with increasing difficulty, and is used to train MπNets to generate smooth motions from depth camera observations. It has a scale of 3.27 million problems distributed across more than 575,000 environments, and the task aims to generate collision-free motions.