Improving Enemy Intelligence in 3D Games

My thesis focuses on improving enemy intelligence in 3D games. The development of reactive yet unpredictable agents is vital to the creation of interactive and immersive gameplay. I attempted to achieve this through two approaches: using a machine-learning model and integrating fuzzy logic to simulate enemy personalities. The machine learning model I developed aimed to create adaptive agents that learn from their environment, while the fuzzy logic state machine adds variance to enemy behaviors, creating more challenging opponents. My machine-learning approach involved the implementation of a Python-based machine-learning package within the Unity game engine to simulate the learning of various games. Fuzzy logic was integrated by giving each instance of an enemy a personality matrix that governs the flow of their state machine. I encountered a variety of problems when trying to train my machine-learning model but was still able to learn about the potential applications. My work with fuzzy logic showed great promise in creating a better gaming experience for players through more dynamic enemies. I conclude by emphasizing the potential of these approaches to enhance the gaming experience and the importance of continued research in improving enemy intelligence.