Abstract
Current tactical formations used in real-time strategy games are mainly fixed. However, fixed formations are incapable of adaptation to unforeseen game situations. This paper proposes the emergent tactical formation based on genetic algorithm (GA) to address this issue. The proposed method first parameterizes tactical formations and then uses GA to optimize the parameters. The experimental results on StarCraft testbed show that our emergent tactical formation can defeat the intrinsic game AI in the testbed of small flat-terrain environment, which validates the effectiveness of the proposed method. © 2011 IEEE.