Swarm intelligence facts for kids

Swarm intelligence (often called SI) is a cool idea used in artificial intelligence. It was first talked about by Gerardo Beni and Jing Wang in 1989, when they were working on robot systems.

Imagine a group of simple "agents" or "boids" (like tiny robots or computer programs) that talk to each other and their surroundings. They follow very simple rules. Even though no one is telling each agent what to do, their small, often random, interactions lead to "smart" group behavior. This smart behavior isn't known by any single agent!

Think of nature for inspiration:

• Ant colonies
• Bee colonies
• Bird flocking
• Fish schooling

When we use swarm ideas for robots, it's called swarm robotics. But swarm intelligence is a bigger idea that includes many different computer methods. We can even use "swarm prediction" to guess future events.

How Swarms Behave

Boids: How Birds Flock

Boids is a computer program that acts like real life. It was made by Craig Reynolds in 1986 to show how birds flock. The name "boid" is short for "bird-oid object," meaning something that acts like a bird.

Just like in nature, the complex behavior of Boids comes from simple rules that each "boid" follows. Here are the main rules:

• Separation: Each boid tries to avoid bumping into its neighbors.
• Alignment: Each boid tries to fly in the same direction as its neighbors.
• Cohesion: Each boid tries to stay close to the center of its group.

You can add more rules, like avoiding obstacles or flying towards a goal.

Self-Propelled Particles

Self-propelled particles (SPP), also known as the Vicsek model, were introduced in 1995 by Vicsek and his team. This model is like a simpler version of Boids. In SPP, particles move at a steady speed. But they change their direction to match the average direction of other particles nearby.

SPP models suggest that all swarming animals, no matter what kind, share some group behaviors. Swarming systems create "emergent behaviors" – meaning new, complex actions appear from simple rules. These behaviors happen at many different levels and are often very strong and common. Scientists are trying to find simple computer models that can show these behaviors.

Swarm Algorithms

Scientists use ideas from nature to create smart computer methods called metaheuristics. Some popular ones include particle swarm optimization (PSO) and ant colony optimization (ACO). These methods are good at finding solutions to hard problems, even if they don't always guarantee the *perfect* answer.

Ant Colony Optimization

Ant colony optimization (ACO) was created by Marco Dorigo. It's a type of optimization method that copies how an ant colony works. ACO is a smart way to find the best paths through networks or graphs.

Real ants drop pheromones (chemical trails) to show other ants where to find food. In ACO, artificial "ants" (computer programs) move around a problem space. They record their paths and how good their solutions are. Over time, more "ants" follow the paths that led to better solutions, helping the system find the best answer.

Particle Swarm Optimization

Particle swarm optimization (PSO) is another smart method that finds the best solution in a multi-dimensional space. Imagine a group of "particles" (like tiny search agents) moving around this space. Each particle has a starting speed and can "talk" to other particles.

After each step, the particles check how good their current solution is. They then speed up towards other particles in their group that have found better solutions. A big plus of PSO is that it has many particles, which makes it good at avoiding getting stuck in "local minima" (solutions that look good but aren't the absolute best).

Artificial Swarm Intelligence (Human Swarming)

Artificial Swarm Intelligence (ASI), also known as Human Swarming or Swarm AI, started in 2015. It's a way to make human groups smarter by using computer rules inspired by natural swarms. This technology connects groups of people online in real-time. They work together like a dynamic swarm to discuss and agree on answers to questions.

ASI has been used for many things:

• Helping business teams make very accurate financial guesses.
• Allowing sports fans to predict game outcomes better than experts.
• Helping groups of doctors make much more accurate diagnoses. For example, a study in 2018 showed that groups of doctors using ASI could diagnose medical conditions with much higher accuracy than individual doctors. Another study in 2021 showed a 23% increase in accuracy for diagnosing MRI images.
• The Food and Agriculture Organization (FAO) of the United Nations has used ASI to help predict famines around the world.

What Swarm Intelligence is Used For

Swarm intelligence methods can be used in many ways.

• The U.S. military is looking into using swarm ideas to control unmanned vehicles (like drones).
• The European Space Agency is thinking about using orbiting swarms for building things in space.
• NASA is exploring swarm technology for mapping planets.
• Scientists have even thought about using swarm intelligence to control tiny robots (nanobots) inside the body to fight cancer.
• Swarm intelligence is also used for data mining (finding patterns in large amounts of data) and cluster analysis (grouping similar things together).

Ant-Based Routing

Swarm intelligence has also been studied for use in telecommunication networks, especially for "ant-based routing." This idea was first explored in the mid-1990s.

It works like this: small "ant" packets travel through a network. They leave a "pheromone" trail on the paths they successfully use. This helps other packets choose the best routes. Airlines have even used ant-based routing to assign aircraft to airport gates. At Southwest Airlines, a computer program uses swarm theory. Each pilot acts like an ant looking for the best gate. The program helps pilots find gates they can arrive at and leave from quickly, which helps the whole airline.

Crowd Simulation

Artists use swarm technology to create complex interactive systems or to simulate crowds in movies and games.

Examples in Movies

• The Lord of the Rings movies used similar technology called Massive (software) for huge battle scenes. Swarm technology is popular because it's affordable, strong, and simple.
• Stanley and Stella in: Breaking the Ice was the first movie to use swarm technology to show realistic movements of fish and birds using the Boids system.
• Tim Burton's Batman Returns also used swarm technology for showing how groups of bats moved.

Airlines have also used swarm theory to simulate how passengers board a plane. This helps them find the fastest ways to get people onto flights.

Human Swarming

Networks of people can be organized into "human swarms" using real-time computer systems. This idea, developed by Louis Rosenberg in 2015, lets the combined intelligence of connected groups of people online be used. The group's combined intelligence is often better than what any single person in the group could do alone.

For example, Stanford University School of Medicine published a study in 2018. It showed that groups of human doctors, when connected by real-time swarming methods, could diagnose medical problems much more accurately than individual doctors or groups using older methods. In one study, swarms of radiologists (doctors who read X-rays) reduced errors by 33% compared to traditional methods.

The University of California San Francisco (UCSF) School of Medicine also found in 2021 that small groups of doctors working together with Artificial Swarm Intelligence (ASI) technology had a 23% increase in accuracy when diagnosing MRI images.

Swarmic Art

Scientists have used swarm intelligence algorithms to create new kinds of art. They use methods that copy how ants forage for food or how birds flock. These "artist" swarms can create unique drawings and paintings from input images. They explore the digital canvas and focus on detailed areas, making each artwork different every time.

Notable Researchers

Images for kids

• Ant colony optimization example.gif

  An example of Ant Colony Optimization, showing how artificial ants find the shortest path.

• Boids-screenshot.png

  A screenshot of a Boids simulation, showing how simple rules can create complex flocking behavior.

See also

In Spanish: Inteligencia de enjambre para niños

• Cellular automaton
• Complex systems
• Language
• Reinforcement learning