Swarm Intelligence

Intelligent group behavior that emerges when many simple individuals interact only through local rules.

Swarm intelligence refers to the collective behavior shown by systems made of many simple individuals with no central controller: each individual senses only its neighbors and follows simple rules, yet the group self-organizes into complex outcomes such as foraging, nest-building, or flocking. Ant colonies, bee hives, bird flocks, and fish schools are all natural examples. The term was coined by Beni and Wang in 1989 while studying cellular robotic systems. It gave rise to algorithms such as ant colony optimization and particle swarm optimization, and to swarm robotics: using large numbers of low-cost robots that cooperate on search, coverage, transport, and formation tasks. The approach scales easily, and losing an individual robot does not compromise the whole. It is closely related to multi-robot collaboration and multi-agent reinforcement learning.

ExampleIn a Harvard experiment published in Science in 2014, 1,024 Kilobots, each just a few centimeters across, self-assembled into shapes such as a five-pointed star and the letter K using only infrared communication with their neighbors.

Also called

SI, Collective Intelligence, Swarm Robotics

Related

Multi-robot Collaboration · Multi-Agent Reinforcement Learning · Emergent Abilities · Multi-Agent Path Finding · Unmanned Aerial Vehicle (UAV)

Sources

Swarm intelligence - Wikipedia
A self-organizing thousand-robot swarm (Harvard SEAS, 2014)