Python wrapper for the maze navigation evolutionary robotics domain common across many of my publications.
The program is built using SCons (http://www.scons.org/)
You'll need to download these to run the maze experiments.
- Hard maze data file: https://s3.amazonaws.com/precomputed-maze/hard_maze.zip
- Medium maze data file: https://s3.amazonaws.com/precomputed-maze/medium_maze.zip
Python dependencies:
- Numpy
- Numba (http://numba.pydata.org/)
- Scipy
Download one or both of the precomputed maze domain files (hard_maze.zip and/or medium_maze.zip) and unzip them into the logs subdirectory.
Running demo.py by default will initiate a run of fitness-based search in the medium maze.
The file demo_notebook.ipynb demonstrates some basic analysis of the data.