MAgent Grid-World

该数据集提供了一个格子世界平台，用于多智能体环境中的协作学习。在这些环境中，智能体通过各种场景，如战斗、丛林和路径规划，学会相互合作。该环境模拟了具有局部观察和离散动作（如移动或攻击）的智能体。规模上，数据集包含了30个智能体和12个敌人。任务是在动态环境中实现多智能体协作。

This dataset provides a grid-world platform designed for collaborative learning in multi-agent environments. In these environments, agents learn to cooperate with each other through various scenarios including combat, jungle exploration, and path planning. The environment simulates agents with partial observable states and discrete action spaces, such as movement and attack. Regarding the scale, the dataset includes 30 agents and 12 enemy units. The core task of this platform is to enable multi-agent collaboration in dynamic environments.