Automatic Strategies for Autonomous Virtual Characters

Abstract

Technique presented in the paper concerns automatic generation of strategies controlling virtual characters’ behaviour. Virtual people are currently widely used in many applications, especially in computer games, films and educational systems. A lot of researches focus on creating intelligent characters capable of deciding about their actions. The fully acceptable solution has not yet been found. The paper presents the problem of generating strategies by means of modified genetic programming. A new Guide-Path technique is introduced.