Portal de Periódicos da CAPES

Gold rush - A swarm dynamics in games

2017; American Institute of Physics; Volume: 1863; Linguagem: Inglês

10.1063/1.4992253

1935-0465

Ivan Zelinka, Michal Bukáček,

Reinforcement Learning in Robotics

This paper is focused on swarm intelligence techniques and its practical use in computer games. The aim is to show how a swarm dynamics can be generated by multiplayer game, then recorded, analyzed and eventually controlled. In this paper we also discuss possibility to use swarm intelligence instead of game players. Based on our previous experiments two games, using swarm algorithms are mentioned briefly here. The first one is strategy game StarCraft: Brood War, and TicTacToe in which SOMA algorithm has also take a role of player against human player. Open research reported here has shown potential benefit of swarm computation in the field of strategy games and players strategy based on swarm behavior record and analysis. We propose new game called Gold Rush as an experimental environment for human or artificial swarm behavior and consequent analysis.