Evolved creative intelligence for computer generated forces
Linus J. Luotsinen, Farzad Kamrani, Peter Hammar, Magnus Jändel, Rikke Amilde Løvlid · 2016
This paper provides an example of using genetic programming for engendering computational creativity in computer generated forces, i.e. simulated entities used to represent own, opponent and neutral forces in military training or decision support applications. We envision that applying computational creativity in the development of computer generated forces may not only reduce development costs but also offer more interesting and challenging training environments. In this work we provide experimental results to strengthen our arguments using a predator/prey game. We show that predator behavior created by a computer, using genetic programming, surpasses predator behavior manually programmed by humans and argue that the sparse automatically generated code is unlikely to be generated by a human and therefore can be considered as a good example of computational creativity. Although the experiments are not conducted in a real-world training simulator they provide valuable insight that exemplifies the opportunities and the challenges of computational creativity applied to computer generated forces.