Graf neuronske mreže za primjenu u višerobotskim sustavima

undergraduate thesis

• Title (english): Graph Neural Networks for Application in Multi-Robot Systems
• Author: Mia Sara Matušin
• Mentor: Tamara Petrović (mentor)
• Committee member: Tamara Petrović (član povjerenstva)
• Committee member: Marko Križmančić (član povjerenstva)
• Committee member: Marijana Peti (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2024-07-12, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Ovaj rad proučava graf neuronske mreže, posebnu vrstu neuronske mreže za učenje na podacima koji se mogu prikazati grafom. Graf neuronske mreže su implementirane i testirane za dva slučaja uporabe. Prvo se testira primjena graf neuronskih mreža u estimaciji Fiedlerove vrijednosti grafa, a rezultati pokazuju točnost između 50-85%. Drugo, istražuje se primjena na višerobotske sustave. Za takve sustave koji se koriste u skladištima, mapa puta može se prikazati grafom. U ovom radu, za zadan broj vozila i njihove putanje, pomoću graf neuronskih mreža procjenjuje se prosječan broj vozila na svakom bridu. Točnost ostvarena za odabrani skup podataka je 95%. Metoda pokazuje obećavajuće rezultate koji se mogu koristiti za razmatranu ili sličnu primjenu višerobotskih
sustava.

Abstract (english)

This work studies graph neural networks, a special type of neural network for learning on data that can be represented by a graph. Graph neural networks are implemented and tested for two use cases. First, the application of graph neural networks in the estimation of Fiedler value of a graph is tested, and the results show an accuracy between 50-85%. Secondly, the application to multi-robot systems is investigated. For such a system used in warehouses, the roadmap can be represented by a graph. In this work, the average number of vehicles on each edge of such a roadmap graph is estimated using graph neural networks, when a certain number of vehicles and their paths are given. The accuracy on the given dataset was 95%. The method shows promising results that could be used in the planning of this or similar multi-robot applications.

• Keywords: graf neuronske mreže; Fiedlerova vrijednost; višerobotski sustav
• Keywords (english): graph neural networks; Fiedler’s value; multi-robot systems
• Language: croatian
• URN:NBN: urn:nbn:hr:168:941608
• Study programme: Title: Computing; Study programme type: university; Study level: undergraduate; Academic / professional title: sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva (sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2024-10-12 21:32:48