Gameplay Automation

Reinforcement Learning has performed remarkably well in various games and has the potential to surpass human level gameplay. Although it does good in turn-based games, complex game genres like fighting or much more complex 3D shooters are still a challenge. The created ML model can teach itself how to play a specific game so that it can be used to test the games for completeness, bugs, and irregularities. The model is also used to find out if there are any ways to speed run games by exploiting certain in-game mechanisms or to check if any characters or abilities are overpowered. Models were created for a few games to identify how well these AI models can perform and see what kind of differences were required to switch between the games.