Self-Generated Simulation Dataset

该数据集是为了在模拟环境中训练多层次物体重新排列任务的结构概念学习模型而生成的。它包含了具有直观结构的场景，这使得模型能够泛化到包含任意数量物体和更高复杂度的未见场景。该数据集的任务是进行多层次物体重新排列规划。

This dataset is generated for training structural concept learning models on multi-level object rearrangement tasks in simulated environments. It includes scenarios with intuitive structures, empowering the models to generalize to unseen scenarios with arbitrary numbers of objects and higher complexity. The core task of this dataset is multi-level object rearrangement planning.