Učenje robota putem višemodalne interakcije

Vitez, Nikola (2017) Učenje robota putem višemodalne interakcije. = Teaching robot by multimodal interaction. Master's thesis (Bologna) , Sveučilište u Zagrebu, Fakultet strojarstva i brodogradnje, UNSPECIFIED. Mentor: Jerbić, Bojan.

[img]
Preview
Text
Vitez_2017_diplomski.pdf - Published Version Jezik dokumenta:Croatian

Download (2MB) | Preview
[img] Text
Vitez_Nikola_autorska_izjava_diplomski_2017.pdf - Published Version
Restricted to Repository staff only Jezik dokumenta:Croatian

Download (513kB)

Abstract (Croatian)

Ovaj diplomski rad istražuje mogućnosti učenja robota višemodalnom interakcijom čovjeka i robota koja podrazumijeva robotsku percepciju stanja okoline temeljem više vrsta senzora. U radu je najprije opisan korišteni eksperimentalni postav, a zatim je ispitana mogućnost klasifikacije i prepoznavanja radnji pisanja i brisanja ploče. Klasifikacija radnji provedena je na temelju analize mjerenih podataka koji se prikupljaju iz inercijskog mjernog uređaja i vizijskog sustava Polaris Vicra. U drugom dijelu rada je razvijeno učenje robota brisanju na temelju prepoznavanja radnje brisanja razvijenim klasifikacijskim algoritmom te je implementiran primjer rada robotskog sustava na temelju jednostavne klasifikacije naučenih pokreta brisanja.

Abstract

This master thesis explores the possibilities of robot learning by the multi-modal interaction of man and robot, which implies a robotic perception of the environment on the basis of several different types of sensors. In this paper, the experimental setup used was first described, and then the possibility of classifying and recognizing actions of writing and wiping off of whiteboard was examined. The classification of the actions was carried out on the basis of the analysis of measured data collected from the inertial measurement unit and the vision system Polaris Vicra. In the second part of the paper, learning a robot to wipe off the whiteboard has been developed based on the recognition of the action of wiping by a developed classification algorithm, futhermore an example of work of robotic system based on a simple classification of the learned movements during action of wiping is implemented.

Item Type: Thesis (Master's thesis (Bologna))
Uncontrolled Keywords: robot; učenje; klasifikacija; inercijski mjerni uređaj; vizijski sustav
Keywords (Croatian): robot; learning; classification; inertial measurement unit; vision system
Subjects: TECHNICAL SCIENCE > Mechanical Engineering
Divisions: 900 Department of Robotics and Production System Automation > 920 Chair of Manufacturing and Assembly System Planning
Date Deposited: 13 Jul 2017 06:39
Last Modified: 18 Mar 2020 12:34
URI: http://repozitorij.fsb.hr/id/eprint/7774

Actions (login required)

View Item View Item

Downloads

Downloads per month over past year