Here is a short description of how robogen selects parents, creates children and performs evolution.
Some of the
parameters that one can choose are
mu – the number of individuals from which to create children
lambda – the number of new individuals to create in each generation
replacement type – mu,lambda (comma) or mu+lambda (plus)
tournament size
mutation only or mutation and one-point crossover
In the
first generation, the population consists of mu randomly created individuals.
These individuals are put in the simulator and their fitness is evaluated.
From this population, new individuals are created in one of two ways.
- If only mutation is used, a parent is identified using tournament selection and a mutated version is created.
- If both mutation and crossover are used, two parents are identified through tournament selection. They are first crossed over and then mutated to create two new offspring.
This process is repeated till lambda new individuals are created.
The replacement strategy is now employed.
- In the case of mu,lambda (comma), the existing population is replaced by the lambda newly created individuals.
- In the case of mu+lambda (plus), the newly created individuals are added to the existing population.
The individuals are then evaluated for their fitness.
From this set of individuals, the best mu individuals based on fitness are chosen to start the second generation.
This process repeats for the chosen number of generations.
