L11 - Swarm Intelligence: Flocking Behaviour Flashcards
Define what is meant by swam behaviour?
Behaviour which individuals of the swarm can’t perform or at least can’t perform optimally. But as a group, the behaviour can be performed.
Explain the Slime-Mold example of swarm behaviour…
An amoeba that is attracted to and eats matter, but can’t scavenge effectively on its own. If it can’t find food, it splits and duplicates into a collective being (Swarm) that can find food more effectively.
What are some of the types of collective movements?
- Flocking / Shoaling
- Swarms
- Formation travelling
Define Natural Movements and Artificial Movements…
Natural -> 2D and 3D movements that represent the natural world.
Artificial -> 1D movement ( can only move forward and backward ). E.g a train.
What is emergent behaviour?
One of the main properties of flocking in which the movement of the flock naturally converges to be one.
What is Multi-Agent Movements? If it flocking?
A scenario in which many agents are moving similarly, but are not necessarily flocking. For example robots moving in a pre-defined route.
Give the characteristics of flocking…
- No leader
- Dynamic movement as a flock
- Reactive to predators and obstacles
- No collision between flock members
- Tolerant of movement within flocks
What are the 2 ways a flock might react to predators?
Flash Expansion -> Swarm disperses in different directions.
Fountain Effect -> Swarms splits and travels around the predator.
Give some benefits of flocking…
- Increase speed
- Increase range
- Reduce turbulence for back of flock.
- Can confuse predators
Groups migration reduces directional error of the flock. For a flock of N birds, what is the formula to calculate the flock error?
1 / sqrt(N)
What is a Boid? What are it’s properties?
A digital representation of a member of a flock.
- Has direction
- Local awareness -> Only interested in an actions of neighbouring boids.
- Blind spot -> All have a blind spot to represent real animals.
- Identical -> All boids are identical.
Who created Boids? What was the motivation?
Craig Reynolds ( 1987 )
A way to digitally simulate flocking in a computationally cheap way.
What were Reynolds rules for flocking?
Separation -> Steer boids to avoid neighbours
Alignment -> Steer towards average heading of flock mates.
Cohesion -> Steer towards average position of flock mates.
What are the characteristics of Reynolds Flock?
- Synchronise direction changes
- Flocks merge if they meet
- Flash expansion occurs if flocks start together
- If started too far apart, multiple flocks occur which will slowly converge into one