Combat Models for RTS Games

TL;DR

This paper introduces three combat models for RTS games, specifically StarCraft, that simulate combat outcomes and can be learned from replay data to aid tactical decision-making in the absence of explicit forward models.

Contribution

The paper presents novel combat models for RTS games and demonstrates how they can be learned from replay data to improve tactical decision-making.

Findings

Combat models accurately predict combat outcomes.

Using models improves tactical decision quality.

Models can be learned from replay data.

Abstract

Game tree search algorithms, such as Monte Carlo Tree Search (MCTS), require access to a forward model (or "simulator") of the game at hand. However, in some games such forward model is not readily available. This paper presents three forward models for two-player attrition games, which we call "combat models", and show how they can be used to simulate combat in RTS games. We also show how these combat models can be learned from replay data. We use StarCraft as our application domain. We report experiments comparing our combat models predicting a combat output and their impact when used for tactical decisions during a real game.