A Javascript web application to explore and experiment with evolutionary computation principles.
Simulation is comprised of a population of robots with a single wheel whose rotation speed and direction are controlled by a standard PID controller. Parameters for this controller is to be optimized over generations by the genetic algorithm in order to ensure their survival in a challenging environment (bumpy wheel with increasing rotation speed).
Developed by @theo_champion as part of the evolutionary computing course at RMIT university.
- Source code available at https://github.com/theochampion/genetic-playground
- Demo available at https://theochampion.github.io/genetic-playground/
- Report paper available at https://github.com/theochampion/genetic-playground/blob/master/rmit_report_paper_theo.pdf
| Name | Default | Range/Type | Description |
|---|---|---|---|
| Population Size | 100 | Real Number | Size of the population |
| Crossover Rate | 0.6 | [0.0, 1.0] | Probability of crossover operation (multi point) |
| Mutation Rate | 0.1 | [0.0, 1.0] | Probability of mutation operation (safe random bit flip) |
| Individual Selectors | Description |
|---|---|
| IndividualSelectors.Tournament2 | Select fittest of two random individuals |
| IndividualSelectors.Tournament3 | Select fittest of three random individuals |
| IndividualSelectors.Fittest | Select fittest individual of the population |
| IndividualSelectors.Random | Select a random individual |
| Pair-wise (parents) Selectors | Description |
|---|---|
| ParentsSelectors.Tournament2 | Select a pair of individual using IndividualSelectors.Tournament2 |
| ParentsSelectors.Tournament3 | Select a pair of individual using IndividualSelectors.Tournament3 |
| ParentsSelectors.Random | Select a pair of random individuals |
| ParentsSelectors.FittestRandom | Select a pair of the fittest and a random individual |