20.500.12556/DKUM-99414 Razvoj avtonomne mobilne robotske platforme za interno logistiko z optimizacijo poti na osnovi genetskih algoritmov Development of an autonomous mobile robotic platform for internal logistics with path optimization based on genetic algorithms V magistrskem delu najprej obravnavamo teoretične osnove avtonomno vodenih vozil in pripadajočih algoritmov za njihove delovanje, programskega okolja ROS 2 in hevrističnih ter metahevrističnih metod s poudarkom na genetskih algoritmih. Na podlagi teoretičnih osnov je opisana zasnova in implementacija avtonomne mobilne robotske platforme, ki zajema mehansko konstrukcijo, izbiro in integracijo senzorskih sklopov, razvoj programske arhitekture znotraj programskega okolja ROS 2 ter razvoj lastnega algoritma za namene optimizacije poti v skladiščnih okoljih, temelječega na na genetskih algoritmih. Implementirani algoritem vključuje metodo inteligentne inicializacije na podlagi Hammingove razdalje, namenjeno formiranju genetsko raznolike začetne populacije, s čimer se zmanjša verjetnost prezgodnje konvergence v lokalni optimum. Eksperimentalna evalvacija je pokazala, da predlagana metoda v primerjavi s klasično naključno inicializacijo dosega statistično značilno večjo raznolikost začetne populacije ter primerljivo oziroma nekoliko boljšo kakovost končnih rešitev že pri bistveno manjši velikosti populacije, kar je še posebej pomembno za uporabo na mobilnih robotskih platformah z omejenimi računskimi viri. In this master's thesis, we first cover the theoretical foundations of autonomous guided vehicles and the algorithms behind their operation, the ROS 2 software environment, and heuristic and metaheuristic methods with an emphasis on genetic algorithms. Based on the theoretical foundation, the design and implementation of an autonomous mobile robotic platform is described, encompassing mechanical construction, the selection and integration of sensor modules, the development of software architecture within the ROS 2 environment, as well as the development of a custom algorithm for path optimization in warehouse environments based on genetic algorithms. The implemented algorithm includes a method of smart initialization based on Hamming distance, aimed at creating a genetically diverse initial population, which reduces the likelihood of premature convergence to a local optimum. Experimental evaluation has shown that, compared to classic random initialization, the proposed method achieves statistically significantly greater diversity in the initial population and comparable or slightly better quality of final solutions even with a much smaller population size, which is especially important for use on mobile robotic platforms with limited computing resources. Avtonomno vodena vozila Avtomatizirana interna logistika Robot Operating System (ROS 2) Genetski algoritmi Optimizacija poti Autonomous guided vehicles Automated internal logistics Robot Operating System (ROS 2) Genetic Algorithms Path optimization true true false Slovenski jezik Angleški jezik Magistrsko delo/naloga 2026-08-16 14:24:45 2026-09-18 15:00:21 2026-09-19 03:06:29 0000-00-00 00:00:00 2026 0 Maribor 0 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 MAG_Breznikar_Ziga_2026.pdf MAG_Breznikar_Ziga_2026.pdf 1 F8CF97685BA6653A7883A283DBAA6845 5d62fbd826496797ce32196b3378c9caec3bf2aa9a8f17981550635017ecdd3d 75431863-b33d-11f1-a766-00155d0001be https://dk.um.si/Dokument.php?lang=slv&id=212857 Fakulteta za strojništvo 0 0 0