Project Description

Dissertation (final year project) made while at Keele University. The agents have a set of rays shooting from in front of them to sense the environment and a multilayer perceptron as a brain to decide what to do. They can bite food, each other, or poison and have a limited amount of energy. At set intervals the best individuals in the population reproduce and the worst are removed from the simulation, over time agents will be fitter and fitter, surviving more time.

Image displaying the enclosed environment where the simulation takes place. Image showing the representation of food (in magenta) and poison (in green). Their hitboxes are a circle even though they are displayed as a colored square. Image showing the structure of the multilayer perceptron. Size of the input and hidden layer was varied during experiments. Screenshot of the program used to test the functioning of the multilayer perceptron. Image showing the sensors that the agents used to sense the environment. Number and length of sensors was varied during experiments.
Project-related images

Source Code

The project is Open Source, the repository can be found here.

Technologies Used

Here are the main technologies I used to make this game:

Technology Usage
Haxe + HaxeFlixel Language and Framework in which I programmed the game
Echo Physics Physics library mainly used for raycasting
Visual Studio Code IDE used for writing the code

Comments