Artificial Intelligent Player Character Using Neuroevolution of Augmenting Topologies and Neural Networks

There has been a wide research and development of algorithms and methodologies that have focused on the development of Artificial Intelligence supported applications and game agents to train models in order to clear the game without losing. In context to this area of expertise, this paper proposes a training methodology to augmentvirtual players for the game Flappy Bird, which involves steering clear of thecharacter, in this context, which is a bird, of a series of obstacles. The modelling of the game-scenario has been done to depict theactual game and ensure adequate representation of the problem we tried to solve. We have proposed a neural network structurethat uses Neuroevolution of Augmenting Topologies algorithm, which is a genetic algorithm to learn to play the game. We have also implementedself-built fitness functions for genome evaluation along with further enhancements in the algorithms integrated.