中文 | English

StarCraft Commander AI

StarCraft, one of the most successful and challenging RealTime Strategy (RTS) games, has been the "grand challenge" for AI research. After years of research, we developed StarCraft Commander (SCC), a deep reinforcement learning agent plays StarCraft II full game at top professional level, and also a world-class team of algorithms and engineering.

Technical challenge

Complex strategy game

Agent needs to be able to plan long-term as well as react and adjust to counter strategies. It demands full capabilities of macro economy, tech and upgrade, army compositions, battle tactics, attack and defense timing, micro control of units, etc.

Large decision action space

The decision-making action space of Go is within 361, while StarCraft action space is as many as 10 to the 26th power. The complexity grows exponentially with the growth of multi-dimensional decision factors such as decision timing, instructions, subjects, objects, etc.

Incomplete information game

Different from game such as Go which is complete information, StarCraft has a "fog of war" mechanism. The map and opponent information is not available without scouting and opponent modeling. Different from Texas poker which is also incomplete information game, StarCraft environment is dynamic as the battlefield changes rapidly.

Large-scale Multi-Agent

StarCraft requires the coordination of many units (up to two hundreds), whereas games such as "DOTA2" and "Glory of the King" only require control of five units.

Real-time decision making

In order to be competitive in StarCraft, the agent needs to make decisions in Millisecond level. This imposes a challenge on performance and efficiency of the neural network model and optimization.

Research direction

Deep learning

Large scale neural network models, for processing of spatial, sequential and unit information, auto-regressive, multi-task learning, sequence to sequence modeling, reasoning under incomplete information.

Reinforcement learning

Research on cutting edge distributed reinforcement learning framework to support large-scale parallel training, efficient training algorithms, exploration and self-play mechanism to improve sample efficiency.

Evolutionary learning

Research on multi-agent learning and evolution learning, improve agent robustness and diversity.

Evolution of StarCraft Commander (SCC)

Match replay

On June 21, 2020, inspir.ai held a match “Qiyuan AI vs Top Professional Player Challenge" in Beijing. Those replays vs Time and ToodMing (both are StarCraft professional champions in China), together with some games with grandmaster players, are available in the download link below.

Related Products

Contact Us

For business inquiries, please fill in the contact form and we will get in touch as soon as poosible.

Contact