burgerma / sem_robonav_ec

Seminar Roboternavigation - Evolutionary Computing :)

Demonstrator for an Evolutionary Algorithm with small rocktes trying to evolve towards reaching a marked target

Requirements

Use pip install -r requirements.txt to install requirements

Or manually install:

- pygame 1.9.4
- numpy 1.15.1
- matplotlib 2.2.3

Usage

Run python main.py. The evolutionary algorithm runs with a default parameter set. For changing the parameters and modes please refere to the corresponding section below. The live plot shows the average fitness of the current generation, average fitness of children and current maximal fitness in the population. The algorithm runs for 50 generations. At the end the plot is saved as a pdf file in the program directory.

Key bindings

Key	Function
R	run algorithm
S	simulation only (no drawing)
D	activate drawing
ESC	exit

Parameters and modes

Parameters and modes are set by the modes list in main.py that is handed over to the RocketWorld.
The modes list needs the following format:

modes = [pstr, xbool, xstr, mfloat, sstr]

All possible values are listed below.

Table of modes and parameters

pstr	Parent Selection Mode	sstr	Survivor Selection Mode
'lin'	Ranking Linear	'rpl'	replace worst (GENITOR)
'exp'	Ranking Exponential	'1gen'	age-based one Generation
'fps'	Fitness Proportional Selection

xbool	Crossover	xstr	Crossover Mode
True	apply crossover	'1pt'	1-point crossover
False	no crossover	'npt'	n-point crossover
		'full'	full crossover
		'arith'	simple arithmetic crossover

mfloat	Mutationrate
float	set mutation rate