Prepoznavanje poze ronioca putem tjelesne mreže inercijskih senzora

master's thesis

• Title (english): Recognition of diver pose using body-network of inertial sensors
• Author: Tea Tovarović
• Mentor: Nikola Mišković (mentor)
• Committee member: Nikola Mišković (predsjednik povjerenstva)
• Committee member: Zoran Vukić (član povjerenstva)
• Committee member: Marija Seder (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2016-07-12, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Electrical Engineering

Abstract

Ronioci rade u opasnom okruženju pa je razvoj tehnologija koje mogu povećati sigurnost ronioca iznimno važan. Jedna takva tehnologija je DiverNet, mreža inercijskih senzora koja se postavlja po tijelu ronioca u svrhu trodimenzionalne rekonstrukcije poze ronioca tijekom zarona. U sklopu ovog diplomskog rada razvijena su dva algoritma koja prema mjerenjima dobivenima iz DiverNet-a određuju u kojoj se pozi nalazi ronilac. Na taj se način doprinosi boljoj komunikaciji ronioca i površinske stanice te lakšem praćenju stanja ronioca pod vodom. Razvijen je Dynamic Time Warping algoritam i klasifikator umjetne neuronske mreže. Prepoznavanje poze radi se odvojeno za gornji (položaj ruku u
odnosu na tijelo, plivanje, skidanje regulatora, itd.) i donji dio tijela (stajanje, ležanje, vrste plivanja, hodanje itd.). Prikupljanje podataka izvedeno je na 9 osoba pri čemu su snimane statičke i dinamičke poze tijela. Za svaki od implementiranih algoritama prikazani su konačni rezultati, analizirane prednosti i nedostatci, te moguća poboljšanja.

Abstract (english)

Divers are working in dangerous environment and technology that contributes to their safety is very important. One of them is DiverNet, network of inertial sensors that is placed on diver's body in order to three-dimensionally reconstruct diver's pose during the dive. As part of this thesis two algorithms are developed that can determine which pose diver is doing, using only measurements obtained from DiverNet. In this way it contributes to better communication between diver and surface and easier underwater monitoring of the diver. Dynamic Time Warping and Artificial Neural Network classifier are developed. Pose recognition works separately for upper (hand position in relation to the body, swimming, taking off the regulator, etc.) and lower body (standing, lying down, different swimming styles, walking, etc.). The experiments were carried out with a group of 9 people while both static and dynamic poses were recorded. For each of the implemented algorithms final results, advantages, disadvantages and possible improvements are presented.

• Keywords: DiverNet; klasifikacijski algoritmi; DTW algoritam; umjetne neuronske mreže
• Keywords (english): DiverNet; classification algorithms; Dynamic Time Warping algorithm; Artificial Neural Networks
• Language: croatian
• URN:NBN: urn:nbn:hr:168:862408
• Study programme: Title: Electrical Engineering and Information Technology; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka elektrotehnike i informacijske tehnologije (magistar/magistra inženjer/inženjerka elektrotehnike i informacijske tehnologije)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2018-02-06 14:10:01