Skip to main content
Home
About repository
Contact
EN
HR
Repository of Faculty of Science
Browse
By Author
By Year
By Organizational unit
By Scientific field
By Object type
All documents
Advanced search
Upload
Search Term
Access restricted to students and staff of home institution
master's thesis
13
1
A robot training model based on mind-waves
Ivan Anđelić (2016)
University of Split
University of Split, Faculty of science
Department of Informatics
Cite this item:
https://urn.nsk.hr/urn:nbn:hr:166:546580
Metadata
Title
Model uvježbavanja robota temeljen na moždanim valovima
Author
Ivan Anđelić
Mentor(s)
Saša Mladenović
(thesis advisor)
Abstract
Otkrićem EEG valova početkom dvadesetog stoljeća javlja se ideja korištenja moždanih valova za izravnu komunikaciju s okolinom. U ovom radu se demonstrira korištenje moždanih valova za kontrolu jednostavnog lego robota. Za čitanje EEG valova se koriste MindWave slušalice, koje su bežične i jednostavne za upotrebu. Da bi se očitani podaci mogli koristiti za upravljanje robotom s njima treniramo neuronsku mrežu. Izrađena aplikacija omogućuje svim korisnicima definiranje i treniranje personalzirane neuronske mreže, čime se osigurava prilagodljivost sustava svakom pojedinom korisniku. Nakon treniranja mreže korisniku se prikaže graf sa minimalnom pogreškom tokom treninga i omogućeno mu je testiranje mreže, da vidi koliko uspješno je istrenirana. Nakon toga može korištenjem istrenirane neuronske mreže i MindWave slušalica kontrolirati robota.
Keywords
Brain computer interface
Lego Mindstorms
MindWave
NeuroSky
Neural networks
BCI controller
Backpropagation training
error minimization
Parallel title (English)
A robot training model based on mind-waves
Committee Members
Saša Mladenović
(committee chairperson)
Goran Zaharija
(committee member)
Divna Krpan (committee member)
Granter
University of Split
University of Split, Faculty of science
Lower level organizational units
Department of Informatics
Place
Split
State
Croatia
Scientific field, discipline, subdiscipline
TECHNICAL SCIENCES
Computing
Study programme type
university
Study level
graduate
Study programme
Computer Science; specializations in: major in teaching
Study specialization
major in teaching
Academic title abbreviation
mag. educ. inf.
Genre
master's thesis
Language
Croatian
Defense date
2016-09-23
Parallel abstract (English)
Discovery of EEG brain waves in early 20th century lead to the idea of using EEG for direct communication with the environment. This work demonstrates use of mind waves for control of Lego robot. Tool for reading EEG waves are MindWave wireless headphones. Neural networks are used to translate collected data into action for the robot. Application made for the experiment allows each user to define and train personalized neural network. After training, it displays minimum error graph, calculated during training, to the user. Then user can test the trained network, to see how successful was the training. Then the network can be used with MindWave headphones to control the Lego robot.
Parallel keywords (Croatian)
sučelje mozak-računalo
neuronska mreža
unazadna aktivacija
MindWave
Lego Mindstorms roboti
Resource type
text
Access condition
Access restricted to students and staff of home institution
Terms of use
URN:NBN
https://urn.nsk.hr/urn:nbn:hr:166:546580
Committer
Anđela Gudelj