We created a continuum model for flocking with the underlying hypothesis that collectives minimize some kind of energy functional, akin to action functional in classical mechanics. This approach yields a natural class of optimal control problems. The solutions are in the form of wave equations which describe information transfer in flocks, such as those triggered by predator attacks.
Flight recordings of multiple events of European starling yield time-signals of different energy mode components of the collective, e.g. rigid-body translations, rotations, volume changes, etc. By appealing to techniques from evolutionary game theory, we proposed a notion of cognitive cost that quantifies the cost to flock-scale decision-making (kinetic energy mode distributions).
My analysis of local interactions among a group of mobile agents and implementing them on a robotic test-bed led to several demonstrations of collective behavior, a connection to the Newtonian two-body problem with application in a vision-based robotic path planning problem, and an efficient solution to an optimal control problem for a mobile robot.