Boids Swarm & Emergent AI

• Craig Reynolds' boids model and its origin
• The three simple rules that produce flocking behaviour
• What emergent behaviour means in complex systems
• Steering behaviours and how they extend boids
• Real-world applications in animation, robotics, and simulation

Craig Reynolds' Boids (1986)

In 1986, computer scientist Craig Reynolds created a simulation called Boids (short for "bird-oids") to model the flocking behaviour of birds, schooling of fish, and swarming of insects.

The key insight: realistic flock behaviour arises from each agent following three simple local rules — no central controller, no global plan.

Complex, coordinated group behaviour emerging from simple individual rules is called emergent behaviour.

The Three Rules

Each boid looks at its local neighbourhood (a radius around itself) and applies these rules simultaneously:

Rule 1: Separation

Avoid crowding nearby boids — steer away from neighbours that are too close.

Prevents collisions. Each boid calculates the average direction away from very close neighbours and steers in that direction.

Rule 2: Alignment

Steer toward the average heading of nearby boids.

Makes the flock move together. Each boid matches the average velocity/direction of its local group.

Rule 3: Cohesion

Steer toward the average position of nearby boids (the local centre of mass).

Keeps the flock together. Boids are attracted toward the centre of their local group.

All three rules run simultaneously. The final steering force is a weighted sum of the three:

Steering = w₁ × Separation + w₂ × Alignment + w₃ × Cohesion

Adjusting the weights changes flock behaviour — more cohesion makes tighter flocks; more separation makes looser ones.

Emergent Behaviour

No single boid knows the shape of the flock or is "in charge." The flock forms, splits around obstacles, and reforms — all from purely local interactions.

Emergence: A system-level pattern that arises from interactions between agents, not programmed into any individual agent.

Other examples of emergence in nature:

• Ant trails to food sources
• Termite mound construction
• Traffic jams forming on motorways
• Markets setting prices

Steering Behaviours (Extensions)

Reynolds later extended boids into a library of steering behaviours for individual agents:

These behaviours can be combined with priority or weighted blending.

Applications

Why Boids Matter for AI

Boids demonstrated that intelligence can be distributed — complex, adaptive behaviour does not require a centralised brain or explicit programming of the outcome. This idea underpins:

• Swarm intelligence (ant colony optimisation, particle swarm optimisation)
• Multi-agent systems
• Emergent AI in games

1. State Reynolds' three boids rules in your own words. What global behaviour do they collectively produce?

2. None of the boids knows the overall shape of the flock. Explain how the concept of emergent behaviour accounts for the flock's coordinated movement.

3. The weights assigned to the three rules are changed so that cohesion is very high and separation is very low. Predict what change you would observe in the flock's behaviour.

4. How are boids used in film animation? Give a specific example of a scene type where boids simulation would be more practical than animating each character individually.

5. Explain one similarity and one difference between boids-based swarm intelligence and a genetic algorithm. Both are bio-inspired — what natural phenomenon does each draw from?