Genetic Algorithm

On completion of the Complex Game Systems assessment for AIE, I feel it’s a good time to digest through what I’ve learned. What better way than a blog post!

For my complex game system I choose to write a genetic algorithm, which the only requirement was for it to be used in an application. Using the aieBootstrap framework and Box2D I had a very basic visual representation of the AI Agents, and their goal.

The Genetic Algorithms goal was another object at a random location on screen. Each generation of AI Agents would all spawn in the middle of the screen. Each generation tried solving the problem of finding the quickest path to the goal object. Every agents genome was made up of five chromosomes that represented the moves, either up, down, left or right. So if the goal was screen right, the quickest path would be all right chromosomes in the agents genome.

Originally the task was to be more complex with the goal being a moving target in a grid based movement system. After running into difficulty getting the framework for the Genetic Algorithm to work I was left short of time and had to simplify. Having no prior knowledge or experience with the field of machine learning I underestimated the learning curve for myself.

The project is available on Github for viewing.

*Note: Planning to update blog to include more of a technical breakdown of my implementation with code examples. *

Leave a Reply

Your email address will not be published. Required fields are marked *