Intelligent Android Game using Reinforcement Learning to Change the Enemy's Behavior

Video games like movies and music are considered as a popular entertainment. Because of their intractability, people get the chance to alleviate daily stress, find satisfaction and be the heroic version of themselves. Although 3D games dominate the video game market, 2D games find a specified place for their straightforward and simple gameplay. This paper develops a 2D platformer action game that player should jump on platforms and whether shoot enemies or run from them to end a level. The three unique enemies in the game which have different skills like shooting projectiles, flying, and pursuing the player, use reinforcement learning methods to improve and enhance their decision-making ability according to their previous decisions and results to decide more wisely when facing the player. The novelty of this game is that enemies would be able to change their behaviors and actions according to their current state and the thresholds will be modified at the end of each level based on the player's actions in previous level to improve the effectiveness of the game. Since every player has a specific style of play, the enemies will take different appropriate action towards the different players. The game is developed for Android devices by Unity game engine and C# programming language.