agent starts in lower-left corner and goes towards center. After that it explores potentially better, but unexplored paths. Once optimal path has been found game resets and agent starts in lelf-lower corner and goes by optimal path.

dependencies: pygame numpy

Usage: python tester.py test_maze_01.txt

or with vizualisation:

python tester.py test_maze_01.txt --gui

To show map run:

python showmaze.py test_maze_01.txt

Generate a random maze:

python genmaze.py 12 random_maze_12.txt

to generate dataset on random mazes:

python tester.py random -N 500 --log=WARN --save-path data.txt

Only first go towards center + exploration phase is stored.