A Survey of Real-Time Strategy Game AI Research and Competition in StarCraft

By Santiago Ontañón, Gabriel Synnaeve, Alberto Uriarte, Florian Richoux, David Churchill, Mike Preuss. 2013

463 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Developmental and Educational Psychology, Sociology and Political Science, Reinforcement Learning, Game & Decision Theory

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Developmental and Educational Psychology

In Sociology and Political Science

In Reinforcement Learning

In Game & Decision Theory

Citations in the corpus

Revisiting the Arcade Learning Environment: Evaluation Protocols and Open Problems for General Agents (2017, background): “This is the case in Minecraft (Project Malmo; Johnson et al., 2016), Starcraft (Ontañón et al., 2013), and robotic RL platforms (Sutton et al., 2011).”
Multiagent Bidirectionally-Coordinated Nets: Emergence of Human-level Coordination in Learning to Play StarCraft Combat Games (2017, background): “ts, researchers resort to deep reinforcement learning (DRL) [5, 17, 6] due to the complexity of the environment and action space. For the analysis of complexity of StarCraft combat games, we refer to [29, 30]. One of the key components in using DRL is to learn a communication protocol among agents. ”
TorchCraft: a Library for Machine Learning Research on Real-Time Strategy Games (2016, methodology): “As the scope of this paper is not to give a review of RTS AI research, we refer the reader to these surveys about existing research on RTS and StarCraft AI [16, 17].”
Episodic Exploration for Deep Deterministic Policies: An Application to StarCraft Micromanagement Tasks (2016, background): “For a more detailed overview of AI research on StarCraft, the reader should consult [23].”
The StarCraft Multi-Agent Challenge (2019, background): “…in several communities: work ranging from evolutionary algorithms to tabular RL applied has shown that the game is an excellent testbed for both modelling and planning (Ontanón et al., 2013), however, most have focused on single-agent settings with multiple controllers and classical algorithms.”
StarCraft II: A New Challenge for Reinforcement Learning (2017, background): “[21] and Robertson & Watson [25] for an overview.”
Isaac Gym: High Performance GPU-Based Physics Simulation For Robot Learning (2021, background): “Deep reinforcement learning (Deep RL) has achieved superhuman performance in very challenging tasks, ranging from classic strategy games such as Go and Chess [1], to real-time computer games like StarCraft [2] and DOTA [3].”
TStarBots: Defeating the Cheating Level Builtin AI in StarCraft II in the Full Game (2018, unknown): “Please refer to [18, 19] for a review.”
StarCraft Micromanagement With Reinforcement Learning and Curriculum Transfer Learning (2018, background): “at solving a series of challenges, such as spatial and temporal reasoning, multi-agent collaboration, opponent modeling and adversarial planning [8].”
Hierarchical Decision Making by Generating and Following Natural Language Instructions (2019, unknown): “Learning to play a complete real-time strategy game, including unit building, resources gathering, defence, invasion, scouting, and expansion, remains a challenging problem [15], in particular due to the complexity and variations of commercially successful games (e.”
Evolving effective micro behaviors in RTS game (2014, methodology): “Much work has been done in applying different techniques to designing RTS AI players [5].”
CommAI: Evaluating the first steps towards a useful general AI (2017, methodology): “For example, games such as Go (Silver et al., 2016) and StarCraft (Ontañón et al., 2013) require sophisticated planning skills intuitively associated with intelligence.”
Heuristic Search Techniques for Real-Time Strategy Games (2016, methodology): “Scripted solutions are often used by retail video games and by bots in the starcraft AI competitions [59], with behaviours similar to those implemented by humans in competitive games.”
Mastering Complex Control in MOBA Games with Deep Reinforcement Learning (2019, background): “This work is also related to research on building AI agents for playing StarCraft 1v1 games which have been significantly explored (Ontanón et al. 2013; Robertson and Watson 2014).”
Monte Carlo Tree Search: a review of recent modifications and applications (2021, background): “• huge decision space - Starcraft1, precursor of the genre, has branching factor about 1050 or higher Ontañón et al. (2013), whereas chess about 35 and Go roughly 180,”
Game-Tree Search over High-Level Game States in RTS Games (2014, background): “For example the branching factor in StarCraft can reach numbers between 3050 and 30200 (Ontanón et al. 2013).”
The NetHack Learning Environment (2020, methodology): “Both StarCraft: BroodWar and StarCraft II have been successfully employed as RL environments [64, 69] for research on, for example, planning [52, 49], multi-agent systems [27, 63], imitation learning [70], and model-free reinforcement learning [70].”
Learning macromanagement in starcraft from replays using deep learning (2017, background): “However, bots participating in these competitions rely mainly on hard-coded strategies [6, 20] and are rarely able to adapt to the opponent during the game.”