Inferring the interaction rules governing collective movement of players in field sports

Moving animal groups display the capability to perform highly coordinated maneuvers without a centralized control. Familiar examples from nature that have been studied extensively include flocking behavior in birds, schooling fish and pedestrian movements. It has been demonstrated that complex patterns of collective motion can arise from simple interaction rules through simulation, and it has been hypothesized that these simple rules reflect the processes used by individuals to govern their movement and behavior. While it is unclear to what degree these simulated mechanisms actually reflect those used by different species to control movement, research has shown that a set of relatively simple interaction rules can produce the same global patterns of behavior exhibited in experimental data. The state of group collective motion can be quantified through the use of metrics such as group centroids, measures of group polarization and rotation, along with analysis of the distribution of nearest neighbors (e.g. density plots) to assess the structure of the group in space. Individual level movement patterns are analyzed based on the speed, direction, tangential acceleration and the rate of direction change. Using these measures it is possible to infer a set of interaction rules capable of reproducing the group level dynamics.