Learning Believable Player Movement Patterns from Human Data in a Soccer Game

Player movement patterns are one of the behavioral traits immediately visible to an observer. Thus, a soccer AI system striving for believable (human-like) behavior must ensure the believability of player movements. We show how tracking data of real human players in soccer can be used to create a case-based reasoning AI system, able to simulate realistic player movements in a computer soccer game. Our results are confirmed with a direct comparison of actions made by AI-controlled players and professional athletes.