Detekcija i praćenje značajki prikladnih za upravljanje smjerom kretanja vozila

master's thesis

• Title (english): Detection and tracking of visual features suitable for mobile robot heading control
• Author: Stjepan Vrljičak
• Mentor: Ivan Marković (mentor)
• Committee member: Ivan Marković (predsjednik povjerenstva)
• Committee member: Ivan Petrović (član povjerenstva)
• Committee member: Branimir Novoselnik (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2021-02-19, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Autonomna vožnja vozila je složeni problem čije rješenje obuhvaća niz raznih tehnoloških grana. Računalni vid, kao i duboko učenje, ključni su elementi svakog modela autonomnog vozila. Ovaj rad bavi se implementacijom algoritama monokularne vizualne odometrije za rješavanje problema samostalnog održavanja smjera robota. Pomoću popularne biblioteke OpenCV sa omotačem za programski jezik Python, napravljena je programska podrška kojom se prati putanja vozila. Korekcija putanje vrši se metodama strojnoga učenja i dodatnim pretpostavkama. Modeli dubokog učenja korišteni su za sigurnu (bez sudara) vožnju po zadanoj putanji, a detekcija objekata pomoću neuronskih mreža korištena je za stvaranje posebnih regija interesa koje dalje proširuju funkcionalnost monokularne vizualne odometrije. Za potrebe rada korišten je ručno izrađeni robot pomoću RaspberryPi 4 ploče i kamere bez dodatnih senzora.

Abstract (english)

Creating real-world autonomous vehicles has become one of the most popular problems in modern day computer science and robotics. Over the last decade, computer vision, as well as deep learning, has developed state-of-the-art methods for solving problems related to autonomous driving. In this paper, we discuss the problem of mobile robot heading control and monocular visual odometry (MVO) as one of the core methods for solving the given problem. We've used one of the most popular computer vision libraries OpenCV with its Python wrapper to develop an algorithm that tracks the robots trajectory. Furthermore, using deep learning models we've developed methods for safe driving (driving without collisions) and we've used object detection methods to find region of interests that expand the functionality of MVO. We've used homemade robot based on RaspberryPi 4 and its camera module without any additional sensors for all testing purposes.

• Keywords: vizualna odometrija; duboko učenje; monokularna vizualna odometrija; detekcija značajki; kretanje robota; opencv; tensorflow
• Keywords (english): visual odometry; deep learning; monocular visual odometry; feature detection; robot movement; opencv; tensorflow
• Language: croatian
• URN:NBN: urn:nbn:hr:168:657164
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-01-13 10:52:21