Trabajo Fin de Grado
Controlling videogame elements using pre-trained neural networks in Unity
Autor/es | Bermudo Bayo, Miguel |
Director | Hernández Salmerón, Inmaculada Concepción
Borrego Díaz, Agustín |
Departamento | Universidad de Sevilla. Departamento de Lenguajes y Sistemas Informáticos |
Fecha de publicación | 2020 |
Fecha de depósito | 2021-09-14 |
Titulación | Universidad de Sevilla. Grado en Ingeniería Informática – Ingeniería del Software |
Resumen | With this Project we aspire to answer to the need game developers have to develop AI for their intelligent Agents in a faster, less costly way.
we chose this project as it combines my passion for developing videogame with ... With this Project we aspire to answer to the need game developers have to develop AI for their intelligent Agents in a faster, less costly way. we chose this project as it combines my passion for developing videogame with my curiosity for state-of-the-art technology. We prove how intelligent agents can replace manually written AI and perform jobs with minimal supervision. For this we will be using Proximal Policy Optimization algorithms and machine learning packages provided for our Game Engine of choice, unity in this case. We used SCRUM methodology to guide our development and time scheduling needs. As well as several tools for development such as IDE, source control etc. We’ve concluded in this study that these agents although intelligent are very costly, needing very strong hardware to cover its enormous computing needs, as well as the time taken to train them. |
Cita | Bermudo Bayo, M. (2020). Controlling videogame elements using pre-trained neural networks in Unity. (Trabajo Fin de Grado Inédito). Universidad de Sevilla, Sevilla. |
Ficheros | Tamaño | Formato | Ver | Descripción |
---|---|---|---|---|
Neural Networks In Videogames ... | 4.171Mb | [PDF] | Ver/ | |