Swarm Engineering Flashcards
What is swarm engineering?
Engineering the collective behaviour of a group of agents for a desired purpose
What are the main principles of swarm engineering?
The swarm can solve complex problems that a single individual with simple abilities (computational or physical) could not solve.
The swarm is composed of several/many individuals, some of which may ben lost or make mistakes, but performance is not affected.
Individuals in a swarm have little/no memory; they do not know the global state of the swarm or its goal.
4 key points about the agents in a swarm
They are/work in large numbers They are (relatively) homogeneous They are (relatively) incapable Their interactions are based on simple behavioural rules that exploit only local information
What are the two ways agents can communicate?
Direct interaction (touch, wifi etc) Stigmergy (communication through environment)
What are two main challenges of swarm intelligence?
Finding local rules for desired swarm behaviours
Proving that your system does what it is supposed to do, and doesn’t do what it it isn’t supposed to do.
What are the 3 main rules in Reynolds Flocking?
- Separation: aim is to maintain distance from others
- Cohesion: aim is to move towards centre of mass of neighbours
- Alignment: aim is to align the angle to mimic others in neighbourhood
Three strategies combined in Particle swarm optimisation.
Brave: keep flying in same direction
Conservative: fly back to its own best previous position
Swarm: move towards best neighbour
How do ants collectively find shortest paths?
By secreting pheromones.
As they move ants deposit pheromone
Pheromone decays in time
Ants follow path with highest pheromone concentration
Without pheromone, equal probability of choosing short or long path
Shorter path allows higher number of passages and therefore pheromone level will be higher on shorter path
Ants will increasingly choose shorter path