Exploiting Early Intent Recognition for Competitive Advantage

In physical domains (military or athletic), team behaviors often have an observable spatio-temporal structure, defined by the relative physical positions of team members over time. In this paper, we demonstrate that this structure can be exploited to recognize football plays in the Rush 2008 football simulator. Although events in the simulator are stochastically generated, we present a method for reliably recognizing football plays at a very early stage using multiple support vector machines; moreover, we demonstrate that having this early information about the defense’s intent can be utilized to improve offensive team play. Our system evaluates the competitive advantage of executing a play switch based on the potential of other plays to increase the yardage gained and the similarity of the candidate plays to the current play. Our play switch selection mechanism outperforms both the built-in offense and a greedy yardage-based switching strategy.