Simulated 2D Environment for Source Search

该数据集包含了一个模拟的2D环境，其尺寸为20米乘20米，被划分为一个20乘20的网格单元。在这个环境中，每个单元都有一个可能包含障碍物的概率。环境的复杂度设计旨在挑战源搜索决策的难度。此外，该数据集用于评估源搜索过程的有效性和效率，比较大型模型辅助搜索与基准算法之间的性能。规模上，包含了20个基准场景，每个场景运行10次，任务是在模拟环境中进行源搜索。

This dataset contains a simulated 2D environment with a size of 20 meters × 20 meters, which is divided into a 20×20 grid of cells. Each cell has a probability of harboring obstacles. The environmental complexity is intentionally designed to pose challenges to the decision-making process of source search tasks. Furthermore, this dataset is developed to evaluate the effectiveness and efficiency of source search workflows, by comparing the performance between Large Language Model (LLM)-augmented source search and baseline algorithms. In terms of scale, the dataset includes 20 benchmark scenarios, with each scenario being run 10 times, where the core task is to conduct source search within the simulated environment.