Forward Modeling for Partial Observation Strategy Games - A StarCraft Defogger
This work addresses the challenge of partial observability in strategy games for AI systems, with incremental improvements in bot performance.
The paper tackles the problem of state estimation and future prediction from partial observations in real-time strategy games, specifically StarCraft, by using encoder-decoder neural networks that combine CNNs and RNNs, resulting in improved win rates for a rule-based bot against community bots.
We formulate the problem of defogging as state estimation and future state prediction from previous, partial observations in the context of real-time strategy games. We propose to employ encoder-decoder neural networks for this task, and introduce proxy tasks and baselines for evaluation to assess their ability of capturing basic game rules and high-level dynamics. By combining convolutional neural networks and recurrent networks, we exploit spatial and sequential correlations and train well-performing models on a large dataset of human games of StarCraft: Brood War. Finally, we demonstrate the relevance of our models to downstream tasks by applying them for enemy unit prediction in a state-of-the-art, rule-based StarCraft bot. We observe improvements in win rates against several strong community bots.