Hands-On Neuroevolution with Python

Running the hard-to-solve maze navigation experiment

To run this experiment, we can use the same experiment runner that we developed for the simple maze experiment, with the only difference being that different command-line parameters should be provided at the start. You can start the hard maze experiment with the following command:

$ python maze_experiment.py -m hard -g 500 -t 10 --width 200 --height 200

This command starts the hard-to-solve maze experiment for 10 trials with 500 generations each. The width and height parameters determine the dimensions of the subplot to draw the maze records collected during the experiment.

Using the NEAT-Python library for the hard maze experiment, we were unable to find a successful maze solver within ...