Playing a 3D Tunnel Game Using Reinforcement Learning
|Thesis title in Czech:||Hraní 3D tunelové hry pomocí zpětnovazebního učení|
|Thesis title in English:||Playing a 3D Tunnel Game Using Reinforcement Learning|
|Key words:||tunnel game|reinforcement learning|artificial intelligence|algorithms|
|English key words:||tunnel game|reinforcement learning|artificial intelligence|algorithms|
|Academic year of topic announcement:||2021/2022|
|Type of assignment:||Bachelor's thesis|
|Department:||Department of Software and Computer Science Education (32-KSVI)|
|Supervisor:||Adam Dingle, M.Sc.|
|Author:||hidden - assigned and confirmed by the Study Dept.|
|Date of registration:||14.04.2022|
|Date of assignment:||15.04.2022|
|Confirmed by Study dept. on:||28.04.2022|
|Date of electronic submission:||05.01.2023|
|Tunnel games are a 3D video game genre in which the player advances through a tunnel and attempts to avoid obstacles by rotating around the tunnel. The student will write a tunnel game using the open-source Godot game engine. In this game, the player will avoid traps with a variety of shapes, will be able to shoot several types of creatures, and will need to collect enough energy to avoid running out of it. After that, the student will create agents that attempt to learn to play the game using reinforcement learning techniques. Some of these agents will use discrete tabular methods, and others will use linear or non-linear approximate methods, possibly using an artificial neural network. The input to the agents will be either the internal game state or, possibly, the 3D pixel data visible to the game's player. The student will compare and analyze the performance of these agents.|
|Sutton, R. S., & Barto, A. G. (2018). Reinforcement Learning: An Introduction. MIT Press.
Mnih, V., Kavukcuoglu, K., et al (2015). Human-level control through deep reinforcement learning. Nature, 518(7540), 529-533.
Goodfellow, I., Bengio, Y., & Courville, A. (2016). Deep Learning. MIT Press.
Morales, M. (2020). Grokking Deep Reinforcement Learning. Manning Publications.