Ανάπτυξη αυτόνομου οχήματος με ενισχυτική μάθηση σε περιβάλλον προσομοίωσης
Reinforcement learning - based self driving vehicle in simulation environment
View/
Keywords
Autonomous driving ; Self driving vehicle ; Simulation ; Reinforcement learning ; DDPG ; CARLAAbstract
This master's thesis focuses on the development of an autonomous vehicle in a digital environment, based on advanced reinforcement learning methods. Initially, a literature review of relevant studies concerning the development of autonomous vehicles in digital environments was conducted, aiming to understand modern approaches and technologies. Furthermore, the theoretical background is presented, covering fundamental definitions as well as more specialized concepts and reinforcement learning techniques. The algorithm selected for the implementation of the model is the Deep Deterministic Policy Gradient (DDPG), while the CARLA simulation software was used as the simulation environment. Experiments were carried out to evaluate the model’s performance under various routes and weather conditions. This thesis contributes to a deeper understanding of the challenges related to autonomous vehicle development and proposes future improvements in the field of autonomous driving.
Postgraduate Studies Programme
Προηγμένα Συστήματα ΠληροφορικήςDepartment
Σχολή Τεχνολογιών Πληροφορικής και Επικοινωνιών. Τμήμα ΠληροφορικήςNumber of pages
70Language
GreekCollections
Except where otherwise noted, this item's license is described as
Αναφορά Δημιουργού-Μη Εμπορική Χρήση-Όχι Παράγωγα Έργα 3.0 Ελλάδα
Αναφορά Δημιουργού-Μη Εμπορική Χρήση-Όχι Παράγωγα Έργα 3.0 Ελλάδα