Battlefield Environment Design for Multi-agent Reinforcement Learning

In reinforcement learning, an agent interacts with an environment for learning its policy. Designing the environment is an important part of training the agent because the change of the environment can affect the agent's policy. In this paper, we introduce the new battlefield environment for multi-agent reinforcement learning. In our environment, two groups of agents compete for their contrasting goals with limited information. As a result, we define the map of the environment and design the state, action, reward, and transition of each agent.