Genetic Algorithm

The goal of this project was to train an AI agent capable of efficiently avoiding enemy units within a game level. On a more nuanced level this can be viewed as a simulation of a population of animals avoiding a predator. 

The agents have 3 base stats:

• speed - how fast the agent moves
• vision - the range at which the agent detects the enemy
• intelligence - determines if the agent acknowledges the presense of an enemy when in range

The scene begins an initial population. The initial population has randomly generated stats. A fitness level is calculated for each agent. The selection process comes in the form of the predator killing enemies. After a certain period, the crossover stage beings, where surviving agents breed and produce offspring. A chance for mutation is present during the crossover stage in order to reduce chances of genetic stagnation.