Small and Large Maps for LMAPF

该数据集专为终身多代理路径查找（LMAPF）实验设计，包含小地图和大地图。小地图用于训练，而大地图则用于评估，其障碍物结构各异。数据集还包括对不同代理数量以及其他方法（如Follower）所使用的基准评估。该数据集已在具有挑战性的仓库环境中通过真实机器人进行了验证。规模上，数据集涵盖了多达10,000个代理在六个大规模地图上的情况。任务涉及的是具有终身目标和动态重新规划的多代理路径查找（MAPF）。

This dataset is designed for lifelong multi-agent path finding (LMAPF) experiments, including both small-scale and large-scale maps. Small maps are used for training, while large maps with diverse obstacle structures are reserved for evaluation. The dataset also includes benchmark evaluations for varying numbers of agents and other approaches such as Follower. It has been validated using real robots in challenging warehouse environments. In terms of scale, the dataset covers scenarios with up to 10,000 agents across six large-scale maps. The tasks involved are multi-agent path finding (MAPF) with lifelong goals and dynamic replanning.