Multi-agent Cooperative Control in Neural MMO Environment Based on MAPPO Algorithm

In recent years, with the increasing number of artificial intelligence and deep learning algorithms and applications, reinforcement learning, as a branch of machine learning, has shown its advantages over other machine learning algorithms in highly complex environments, including famous events such as AlphaGo defeating human chess players. Various branches of reinforcement learning algorithms and experimental environments have been developed for research. Reinforcement learning and multi-agent reinforcement learning are applied in this paper. The multi-agent reinforcement learning algorithm MAPPO with a proposed reward function is validated under a Neural MMO environment. The result verifies that the MAPPO algorithm can provide strategies for agents running in the Neural MMO environment.