Subscribe
Subscribe
MY AMERICAN SCIENTIST
LOG IN! REGISTER!
SEARCH
 
Logo IMG

COMPUTING SCIENCE

E Pluribus Unum

Brian Hayes

Boids of a Feather

Bird flocks are another classic application of StarLogo modeling. Starlings and several other bird species assemble in large cluster flocks that perform synchronized, balletic flights, wheeling and swooping over autumn fields. People have long wondered both why and how they do it. For a time the "how" question was considered a problem in animal communication: How can the birds signal their intentions quickly enough to coordinate the group's sudden movements? There was even a suggestion (by Edmund Selous, in a bizarre but charming book published in 1931) that flocking birds must rely on some form of "thought transference" or "collective thinking" for synchronization. It was only in the 1980s that another kind of explanation came into favor. That new idea was inspired in part by computer simulations.

In discussing the history of ideas about bird flocks, Frank Heppner of the University of Rhode Island has written that "a zeitgeist was at large" in the mid-1980s, when several workers independently devised similar theories. Heppner himself was one of the theorists; he "suggested that flocking might be an emergent property arising out of simple rules of movement followed by individuals in the flock." At about the same time Craig W. Reynolds, then at Symbolics Inc., produced a computer simulation of flocking "boids" that relied entirely on local interactions between near neighbors. (The boids attracted attention not only among biologists and computer scientists but also in Hollywood; Reynolds recently received an Academy Award for his work on this and several other animations.)

Reynolds's boids obeyed three main rules: 1) Avoid collisions. 2) Try to match the speed and direction of nearby boids. 3) Move toward the "center of gravity," or the mean position, of the nearby boids. An even simpler flock simulation is included among the sample programs shipped with the StarLogo software. At each time step, each bird finds its nearest neighbor among all its flockmates. (If there are multiple nearest neighbors, all at the same distance, one is chosen arbitrarily.) If the chosen neighbor is closer than a predetermined threshold distance, the bird flies in the opposite direction; if the neighbor is farther than the threshold, the bird flies toward it; if the distance is just right, the bird matches the neighbor's heading. Thus each bird interacts with only one neighbor at a time. This rudimentary model will not win any Oscars, but it creates flocks that do seem recognizably avian.








comments powered by Disqus
 

EMAIL TO A FRIEND :

Of Possible Interest

Computing Science: Clarity in Climate Modeling

Feature Article: Candy Crush's Puzzling Mathematics

Spotlight: New Information from Ancient Genomes

Subscribe to American Scientist