Deep RL Agent for a Real-Time Action Strategy Game
This work addresses the challenge of applying RL to complex real-time games, but it is incremental as it builds on existing methods like PPO and self-play.
The authors tackled the problem of developing a deep reinforcement learning agent for Heroic - Magic Duel, a real-time action strategy game with a large state space and imperfect information, achieving a win rate of around 65% against existing AI and over 50% against a top human player.
We introduce a reinforcement learning environment based on Heroic - Magic Duel, a 1 v 1 action strategy game. This domain is non-trivial for several reasons: it is a real-time game, the state space is large, the information given to the player before and at each step of a match is imperfect, and distribution of actions is dynamic. Our main contribution is a deep reinforcement learning agent playing the game at a competitive level that we trained using PPO and self-play with multiple competing agents, employing only a simple reward of $\pm 1$ depending on the outcome of a single match. Our best self-play agent, obtains around $65\%$ win rate against the existing AI and over $50\%$ win rate against a top human player.