Optimizing NPC Behavior in Video Games Using Unity ML-Agents: A Reinforcement Learning-Based Approach
Sudhanshu Maurya, Arnav Kotiyal, Pankaj Prusty, Nilesh M Shelke, Abhishek Bhattacherjee, Tiyas Sarkar · 2025
The development of video games is now being influenced by machine learning (ML), which is changing the industry with every passing day. The focus of this research is on how video games can be made better through ML in game development, particularly how NPCs behave in the game. Game companies can boost the immerse feeling in video games by utilizing reinforcement learning in developing sophisticated NPCs. Those who play modern video games know that they are more realistic than what they used to be, and this is courtesy of ML. The toolkit Unity ML-Agents enables more straightforward ML implementations in the Unity game engine, giving game makers the ability to create complex AI features without having to have deep knowledge of machine learning. This paper seeks to analyze how ML can transform the experience of humans when playing games by enhancing the NPCs and through interactivity. The future of video games is bright, especially considering new advancements in technology like ML.