DARPA seeks “to develop and apply Artificial Intelligence (AI) to existing open-world video games to quantitatively assess game balance, identify parameters that significantly contribute to balance, and explore new capabilities, tactics, and rule modifications that are most destabilizing to the game.”
Using Starcraft2 as the game, we are developing our novel BREAK technology to achieve these objectives. BREAK leverages our prior work in deep reinforcement learning to build a massive database of game play, perturbations within the state and configuration of the game, and to learn a model of game balance.
Heron Systems is moving beyond the limited objectives of the Gamebreaker program to apply BREAK to create decision aids for acquisition planning, mission planning, and resource allocation tasks.
Massive datasets enable robust learning and ability to explore millions of scenarios.
Generate courses of action with supporting metrics in minutes.
Integrate with any simulation featuring a robust API allowing simulation configuration and agent control.
Targeted domain randomization enables deep exploration of scenarios of interest.