Brainscape's Knowledge GenomeTM


Top Flashcards (Certified)
All Flashcards

Bio-AI > Lecture 8 - Swarms > Flashcards

Lecture 8 - Swarms Flashcards

1
Q

What is swarm engineering?

A

Engineering the collective behaviour of a group of agents for a desired purpose

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

What are 3 positives of swarm engineering?

A
  1. Robust/Flexible
  2. Scalable (though difficult in practice)
  3. More capable than sum of parts
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

What are 5 properties of individual agents in a swarm?

A
  1. Relatively homogeneous
  2. Relatively incapable (as a single agent)
  3. Simple rules define actions
  4. Interact locally only
  5. Little/no memory
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

Name 2 ways agents interact locally.

A
  1. Direct communication (wifi etc)

2. Stigmergy (communication through environment)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

What are the two main challenges of swarm intelligence?

A
  1. Finding local rules that produce desired behaviour

2. Proving your system works

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

What are 4 methods of developing local rules for swarm intelligence?

A
  1. Trial and error
  2. Crowd source (which is essentially trial and error scaled up)
  3. Bio-inspired
  4. Artificial Evolution/Machine Learning
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

What are the 3 main rules of Flocking (Reynolds 1987)?

A
  1. Separation: maintain given distance between each neighbour
  2. Cohesion: move toward the centre of mass of neighbours
  3. Alignment: align angle agent agent faces with neighbours
    (4) /(5): Attractors (Migrator), Repulsor (Object avoidance) - not on slides, said in lecture.
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

What are the 3 basic rules/behaviours of PSO (particle swarm optimisation)?

A
  1. Brave: Explore new areas to find better food sources
  2. Conservative: Fly back to previous best food source
  3. Swarm: Move to best local area (from neighbours)
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

What is particle swarm optimisation?

A

Finding the best ‘food source’ in an area. Optimisation using a swarm to find the highest value target.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

What are the properties of agents in PSO?

A

Can ‘shout’ the value of the food source at their current location to their neighbours
Can remember the location of the best food source it has encountered

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

How are the actions of a PSO agent governed?

A

v(t+1) = av(t) + bR(pbest-S(t)) + cR*(gbest-S(t))

v(t): current velocity
a,b,c: learning rates
R: random number [0,1]
pbest: best position already encountered
gbest: best position from neighbours 'shouts'
s(t): current position
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

What is the main use of ant colony optimisation (ACO)?

A

Find the shortest path between two nodes in a network

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

What is the main advantage of ACO over traditional searching algorithms?

A

Can work in dynamic networks. Traditional search algorithms assume the network is static.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

What is a disadvantage of ACO?

A

Not guaranteed to find the optimal solution in networks >30 nodes

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

What is the basic ACO algorithm?

A
  1. At each node generate a random number
  2. -If rand is less than a threshold value q0: use the transition rule to pick the next path
    - If rand is greater than q0: choose the path with the highest pheromone level
  3. Go along chosen path (edge) to next node. Lay pheromone at same time
  4. Repeat
    Note: (pheromones decay over time)
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

What is the global update for ACO?

A

Once the shortest path has been found, this path is updated to have pheromones increased so inversely proportional to length.

17
Q

What are the two rules for environmental monitoring robots (to follow odor)?

A
  1. If lost: spin until you locate the plume from the odor

2. If following odor: ‘run & tumble’ until you reach the source

18
Q

What are some examples of robot swarm demonstrations?

A

Chaining: using chains of coloured robots to map a path
Coordinated box pushing: (self explanatory)
Blind bulldozing: constructing a ‘nest’ using bulldozers with only local rules
Cooperative manipulation: using pairs of robots to pick up sticks that are too heavy for an individual
Making shapes: Kilobots

19
Q

What are the 2 rules for clustering (objects)?

A
  1. Probability to pick up object: inversely proportional to N (N = number of objects seen in short time window)
  2. Probability to put down object: directly proportional to N
20
Q

Give an example of physical cooperation in swarms.

A

Ants using their bodies to:

  • build bridges
  • cover potholes
  • create nests
21
Q

What is a reconfigurable robot?

A

A modular robot that can be reorganised to create a morphology suitable for different tasks

22
Q

Give 4 examples of structural layouts for reconfigurable robots

A
  1. Lattice
  2. Chain
  3. Mobile
  4. Magnetic
23
Q

What type of communication is used in clustering?

A

Stigmergic

24
Q

What type of communication is used by Kiva Systems (Warehouse organisers)?

A

Global. An example of a ‘swarm’ which is better controlled using global over local communication

Bio-AI (10 decks)

Brainscape helps you reach your goals faster, through stronger study habits.
© 2025 Bold Learning Solutions. Terms and Conditions