Swarm behavior

from Wikipedia, the free encyclopedia
A shoal of Atlantic herring ( Clupea harengus ) migrating to the spawning grounds in the Baltic Sea . The high speed can be maintained over thousands of kilometers. Some scientists believe that moving around in closed groups is energy efficient.

The term swarm behavior describes the behavior of fish , birds , insects or other animals to join together to form aggregations . The individuals in a swarm usually belong to the same species , but mixed swarms of animals of different species and sizes can also form. Typical shoals are herring , starlings and locusts .

The advantages of swarm formation arise in the search for food and in protection against possible predators, e.g. B. through collective vigilance. The individuals of the swarm always move together in one direction. The physiological basis of the ability to move synchronously is believed to be in the mirror neurons .

In the case of land mammals, if they behave similarly, they are referred to as a herd , and if they move faster , they are referred to as a stampede .

The phenomenon is not only studied in various biological disciplines, but also z. B. in computer science (see particle swarm optimization ). There are applications in computer graphics (see particle system ) and in the military.

Rules for the formation of swarms

Bats leave a cave in Thailand just before sunset
Mosquitoes on a treetop
Swarm of bees

Computer simulations of swarms that were modeled for the first time in 1986 by Craig Reynolds brought interesting results . The principle is based on three rules that the individual agents (individuals / boids ) observe:

  1. Move towards the center of those you see around you ( cohesion ).
  2. Move away as soon as someone comes too close to you ( separation ).
  3. Move in roughly the same direction as your neighbors ( alignment ).

As a result of these rules at the individual level, there is an overall structure, namely the swarm. One speaks of emergence .

A group of scientists from the University of Leeds led by Jens Krause researched that a structural memory in schools of fish and birds ensures that a specific school formation is always followed by a very specific one. So they are first arranged in a disordered, chaotic swarm like swarms of mosquitoes and then form a torus .

The physicist and ornithologist Andrea Cavagna from Rome found in an EU research project that birds align themselves with the seven neighboring birds, which is the highest number that birds can differentiate. Generally they keep at least one wing span from one another. When the swarm changes direction, the tip of the swarm does not necessarily react, each individual can cause a change in direction and the whole swarm reorganizes itself as a result. When birds of prey approach, the flock condenses to make targeting an individual more difficult. Sometimes even the bird of prey is so enclosed by the flock that it has to drop away unable to fly. If a bird of prey disturbs the swarm-dissolving approach to the sleeping place, the swarm rises again, often until it gets dark.

Geese at the Müritz

However, many species of migratory birds do not fly in flocks, but in V-shaped trains or, like the cranes , also in long chains at an angle. Computer models for swarm behavior had long failed to compute such V-formations from a swarm of randomly arranged animals flying up from the ground. Valmir Barbosa and Andre Nathan ( Universidade Federal do Rio de Janeiro ) reported in spring 2007 that they had solved the problem: By combining only two requirements for each animal:

  1. Use the lift caused by flapping the wings of a bird flying in front of you.
  2. Take up a position from which you can look ahead undisturbed.

These model calculations were based on flocks of up to 35 animals, and regardless of the original arrangement of these animals, a proper formation always emerged.

Algorithmic Complexity

In an application that simulates swarm behavior, there is no central control for the individual individuals. The next position must be calculated separately for each individual. As a result after the O-Notation an O (n²) - algorithm with a computation time of at n individuals. There are different approaches to keep the computing time constant with a varying number of individuals or at least to reduce the computing time of an O (n²) algorithm.

Even Reynolds tried to deploy for this purpose a 3D grid work in which his Boids be distributed based on their position in the container. Using this latticework, the boids can quickly check the containers in their vicinity for neighbors, which reduces the running time of the algorithm.

Use in the military

The US Air Force began research into an autonomous drone system called LOCAAS ( Low Cost Autonomous Attack System ) in 1998 . This drone system uses an algorithm based on the Craig Reynolds model to fly as a swarm. As soon as up to 192 drones are dropped by a stealth bomber , they begin to communicate electronically and attack enemy troops in swarms.

See also

literature

  • Gabriele Brandstetter , Bettina Brandl-Risi u. Kai van Eikels (Ed.): Swarm (E) Motion. Movement between affect and mass . Rombach, Freiburg 2008 ISBN 3793095002
  • Eva Horn , Lucas Marco Gisi (ed.): Swarms - collectives without a center. A story of knowledge between life and information . Bielefeld: transcript 2009. ISBN 978-3-8376-1133-5

Web links

Individual evidence

  1. ↑ In English there is the special term murmuration for the swarming of starlings
  2. ↑ www.red3d.com Craig Reynolds' page on Boids-Links, Applets, etc. a. (English)
  3. ^ Sandro Mattioli : The unknown flying objects . How swarms organize themselves. Bild der Wissenschaft, issue 3/2009, p. 16
  4. ^ A. Nathan, VC Barbosa: V-like formations in flocks of artificial birds. In: Artificial life. Volume 14, number 2, 2008, pp. 179-188, doi : 10.1162 / artl.2008.14.2.179 , PMID 18331189 . arxiv : cs / 0611032v2 .
  5. ↑ A Model for Simulating Swarm Movement by Craig Reynolds
  6. ^ Low Cost Autonomous Attack System - Global Security
  7. ↑ Explanation about the use of natural swarm behavior in the military (PDF; 1.0 MB)