Moving objects detection and tracking by omnidirectional sensors of a mobile robot

doctoral thesis

• Title (croatian): Otkrivanje i praćenje gibajućih objekata svesmjernim senzorima mobilnoga robota
• Author: Ivan Marković
• Mentor: Ivan Petrović (mentor)
• Committee member: Davor Petrinović (predsjednik povjerenstva)
• Committee member: Mato Baotić (član povjerenstva)
• Committee member: Robert Cupec (član povjerenstva)
• Committee member: François Chaumette (član povjerenstva); strani drzavljanin: Nije dostupno
• Committee member: Patrick Danès (član povjerenstva); strani drzavljanin: Nije dostupno
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; (Department of Control and Computer Engineering); Zagreb
• Defense date and country: 2014, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Electrical Engineering; Automation and Robotics
• Universal decimal classification (UDC): 621.3 - Electrical engineering

Abstract

In mobile robotics measurements from various sensors often yield direction-only information. In this thesis we develop directional statistics based methods for moving object tracking by omnidirectional sensors of a mobile robot. Firstly, the speaker localization problem is solved by modeling the measurements of a microphone array with a convex combination of von Mises distributions and the tracking is solved by Bayesian estimation based solely on the von Mises mixture. Furthermore, in the thesis we analyze the voice activity detection problem from the standpoint of model based voice activity detection methods which are enhanced by supervised learning algorithms. Considering the omnidirectional camera, spherical projection model coupled with displacement information from motor encoders is proposed to segment out static and dynamic features and movement tracking is performed on the sphere with a Bayesian filter based on the von Mises-Fisher distribution. Finally, fusion of heterogenous sensors for object tracking is analyzed in a comparative study of the extended information filter, the unscented information filter and the particle filter.

Abstract (croatian)

U mobilnoj robotici mjerenja različitih senzora često daju informaciju samo o smjeru. U doktorskom radu razvijaju se metode praćenja gibajućih objekata zasnovane na usmjernoj statistici. Problem lokalizacije govornika poljem mikrofona riješen je modeliranjem mjerenja senzora konveksnim zbrojem von Misesovih razdioba te je potom praćenje riješeno Bayesovom estimacijom zasnovanoj samo na zbroju von Misesovih razdioba. U nastavku doktorskog rada razmatra se problem otkrivanja govorne aktivnosti koji je riješen modelsko-statističkim metodama koje su zatim unaprijeđene algoritmima nadziranog učenja. Nadalje, razmatra se problem otkrivanja i praćenja gibajućih objekata svesmjernom kamerom mobilnoga robota. Otkrivanje je riješeno koristeći sferni projekcijski model zajedno s podacima odometrije robota, a praćenje je ostvarno na sferi koristeći Bayesovu estimaciju s von Mises-Fisherovom razdiobom. Na kraju, razmatrana je fuzija heterogenih senzora s ciljem praćenja gibajućih objekata te je urađena komparativna studija proširenog i nederivacijskog informacijskog filtra te čestičnog filtra.

• Keywords: moving object tracking; directional statistics; speaker localization; von Mises distribution; microphone array; voice activity detection; sensor fusion; mobile robotics
• Keywords (croatian): praćenje gibajućih objekata; usmjerna statistika; lokalizacija govornika; von Misesova razdioba; polje mikrofona; otkrivanje govorne aktivnosti; fuzija senzora; mobilna robotika
• Language: english
• URN:NBN: urn:nbn:hr:168:307741
• Study programme: Title: Electrical Engineering and Computing; Study programme type: university; Study level: postgraduate; Academic / professional title: Doktor znanosti elektrotehnike i računarstva (Doktor znanosti elektrotehnike i računarstva)
• Catalog URL: http://lib.fer.hr/cgi-bin/koha/opac-detail.pl?biblionumber=43103
• Type of resource: Text
• Extent: 153 str. ; 30 cm.
• File origin: Born digital
• Access conditions: Open access
• Created on: 2019-04-19 13:11:25