Primerjava algoritmov spodbujevalnega učenja na simulaciji parkiranja avtomobila

KRIŽMAN, KRISTJAN (author), Žabkar, Jure (mentor)

Primerjajte algoritma spodbujevanega učenja DQN in DDPG v danem simulacijskem okolju za parkiranje avtomobila. Znotraj omejitev simulatorja lahko spreminjate opise stanj in akcij tako, da bodo primerni za dana algoritma. Uporabite lahko obstoječe implementacije algoritmov ali razvijete svoje. Poročajte o uspešnosti obeh algoritmov, časovni zahtevnosti in njuni občutljivosti na začetne parametre.

• Language: Slovenian
• Keywords: računalnik, strojno učenje, simulacija, igra, avtomobil, učenje
• Work type: Bachelor thesis/paper
• Typology: 2.11 - Undergraduate Thesis
• Organization: FRI - Faculty of Computer and Information Science
• Year: 2022
• PID: 20.500.12556/RUL-135581
• COBISS.SI-ID: 102689027
• Publication date in RUL: 21.03.2022

KRIŽMAN, KRISTJAN, 2022, Primerjava algoritmov spodbujevalnega učenja na simulaciji parkiranja avtomobila [online]. Bachelor’s thesis. [Accessed 19 May 2025]. Retrieved from: https://repozitorij.uni-lj.si/IzpisGradiva.php?lang=eng&id=135581

Secondary language

• Language: English
• Title: Comparison of two reinforcement learning algorithms in a car parking simulator

Abstract

Compare the DQN and DDPG reinforcement learning algorithms in a given car parking simulation environment. You may change the descriptions of states and actions within the limits of the simulator, to suit the given algorithm. Use existing implementations or develop your own. Report on the performance of both algorithms, their time complexity and their sensitivity to initial parameters.

• Keywords: computer, machine learning, simulation, game, car, učenje