Συμπεριφορά ευφυών πρακτόρων χρησιμοποιώντας ενισχυτική μάθηση μέσω του εργαλείου ML-Agents της Unity
Behavior of intelligent agents using reinforcement learning through Unity's ML-Agent’s toolkit
KeywordsArtificial Intelligence ; Machine learning ; Unity's ML-Agents ; Reinforcement learning ; Intelligent Agents
Game AI is as ancient as AI itself, but with the addition of video games, the discipline has witnessed enormous extension and enrichment in the previous decade. Deep learning development has had a substantial and transformative influence on numerous challenging issues during the last decade, including speech recognition, machine translation, natural language comprehension, and computer vision. Many gaming firms are looking for new solutions and technologies to help their products benefit from AI and Machine Learning. Unity's ML-Agent toolkit, which is also the major component of this thesis, is a very broad and popular tool. There are seven chapters in the paper. The first two chapters discuss the principles of artificial intelligence, gaming AI technologies, machine learning and its structure, as well as some examples of projects created with mlagent's tools. From chapter three to six, we follow the project's development step by step, from requirements through design, coding, and then the simulation demo. Finally, some future considerations and a conclusion manage the thesis's closure.