1 Introduction
2 Methodology
2.1 Simulation architecture
2.2 Scenario
2.3 Swarm supervisor
2.3.1 Actions
2.3.2 Conditions
2.4 Evolving the swarm supervisor
2.4.1 Genetic programming
Node type | Selection choices |
---|---|
Operator | Selector/Sequence (Between 2–5 children) |
Action node | Dispersion/North/South/East/West/North East/North West/South East/South West |
Condition node | \(\mu _x>\)-18, \(\mu _x>\)-16, \(\mu _x>\)-14, ... increment by 2 ..., \(\mu _x>\)14, \(\mu _x>\)16, \(\mu _x>\)18 |
\(\mu _x<\)-18, \(\mu _x<\)-16, \(\mu _x<\)-14, ... increment by 2 ..., \(\mu _x<\)14, \(\mu _x<\)16, \(\mu _x<\)18 | |
\(\mu _y>\)-18, \(\mu _y>\)-16, \(\mu _y>\)-14, ... increment by 2 ..., \(\mu _x>\)14, \(\mu _y>\)16, \(\mu _y>\)18 | |
\(\mu _y<\)-18, \(\mu _y<\)-16, \(\mu _y<\)-14, ... increment by 2 ..., \(\mu _x<\)14, \(\mu _y<\)16, \(\mu _y<\)18 | |
\(\sigma>\)1, \(\sigma>\)2, \(\sigma>\)3, ... increment by 1 ..., \(\sigma>\)14, \(\sigma>\)15, \(\sigma>\)16 | |
\(\sigma<\)1, \(\sigma<\)2, \(\sigma<\)3, ... increment by 1 ..., \(\sigma<\)14, \(\sigma<\)15, \(\sigma<\)16 |
2.4.2 Evolutionary algorithm
Parameter | Value |
---|---|
Population size | 100 |
Test limit | 800 s |
Elitism size | 10 |
Tournament size | 3 |
Single point mutation probability | 0.1 |
Sub-tree growth probability | 0.05 |