Mario Runner AI

About The Project

Infinite platformer runner AI built using NeuroEvolution of Augmenting Topologies (NEAT).

Built With

• Python
• NEAT
• Pygame

Getting Started

There are a few librarier that need to be installed in order for the app to work.

Prerequisites

npm install npm@latest -g

Installation

1. Clone the repo

   git clone https://github.com/andreiec/mario-runner-ai.git

2. Install NPM packages

   npm install

Usage

Mario Runner is built upon NEAT. A very simple to understand workflow of the algorithm is the following

1. Generate the initial population of players with random genomes
2. Since they are built randomly, some players might get further than others
3. Get the highest scoring players and breed their genomes to get the next generation
4. Generate the same amount of players, but with the new and improved genomes
5. Let them run until there are no more players
6. Loop through steps 3-5
   
   

Player Input

There are 3 values used for player input and one for the output (jump button)

• Distance to the next pipe
• Next pipe type
• Y-position of player

User Input

H - Display HUD for generation information
K - Kill all current players and move to next generation

License

Distributed under the MIT License. See LICENSE for more information.