Planiranje robotskog djelovanja zasnovano na tumačenju prostornih struktura

doctoral thesis

• Title (english): Robot behaviour planning based on interpretation of spatial structures
• Author: Marko Švaco
• Mentor: Bojan Jerbić (mentor)
• Committee member: Dubravko Majetić (predsjednik povjerenstva)
• Committee member: Bojan Jerbić (član povjerenstva)
• Committee member: Stjepan Bogdan (član povjerenstva)
• Granter: University of Zagreb; Faculty of Mechanical Engineering and Naval Architecture; Zagreb
• Defense date and country: 2015-04-30, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Mechanical Engineering; Production Mechanical Engineering
• Universal decimal classification (UDC): 004 - Computer science and technology. Computing. Data processing

Abstract

Robot je programabilan mehanizam čije se djelovanje temelji na upravljačkim algoritmima. Prilikom rada u nestrukturiranoj okolini upravljački algoritmi postaju eksplicitne funkcije položaja i vremena u povratnoj vezi sa stanjem okoline. Obradu podataka iz okoline te zaključivanje o odgovarajućem djelovanju robota moguće je temeljiti na principima strojnoga učenja. Predloženo istraživanje bavi se razvojem modela učenja i planiranja djelovanja robota. Proces učenja temelji se na novoj umjetnoj neuronskoj mreži klasifikacijom prostornih struktura. Pojam prostorne strukture podrazumijeva interpretaciju rasporeda poznatih objekata u ravnini koje robot percipira vizijskim sustavom. Umjetna neuronska mreža za klasifikaciju i prepoznavanje prostornih struktura zasniva se na teoriji adaptivne rezonancije. Planiranje djelovanja robota temeljno je na usporednoj evoluciji rješenja razvojem novoga genetskoga algoritma. Genetski algoritam kao osnovni cilj ima prostornu pretvorbu neuređenoga stanja objekata u uređeno. Izvorni znanstveni doprinos rada očituje se u sljedećem: 1) Samoorganizirajuća umjetna neuronska mreža za klasifikaciju i prepoznavanje prostornih struktura zasnovana na teoriji adaptivne rezonancije, koju odlikuje nova dvorazinska klasifikacija po obliku i rasporedu objekata te mehanizam asocijativnoga povezivanja neuređenoga skupa objekata s uređenim i 2) Novi genetski algoritam za planiranje robotskoga djelovanja u nestrukturiranoj radnoj okolini karakteriziran usporednom evolucijskom strategijom za pronalaženje rješenja, s ciljem prostorne pretvorbe neuređenoga stanja objekata u uređeno.

Abstract (english)

A robot can be defined as a programmable mechanism with behaviour based on control algorithms. In nondeterministic environments control algorithms become explicit functions of position with respect to time and feedback of the current environment state. Information processing and reasoning about appropriate robot behaviour can be based on machine learning principles.In the proposed research robot learning and behaviour planning model is developed. Robot learning is based on a new neural network by classifying spatial structures. A spatial structure denotes two-dimensional interpretation of known objects and their arrangement in a plane, acquired by the robot vision system.The neural network for classification of spatial structures is based on adaptive resonance theory. Robot behaviour planning is based on parallel evolution by developing a new genetic algorithm. The main goal of the developed genetic algorithm is spatial transformation of an unordered set of objects into an ordered object set.rnOriginal scientific contribution:rnA selforganizing artificial neural network for recognition and classification of spatial structures based on Adaptive resonance theory, with a novel two-level classification based on object shape and arrangement and a new mechanism for associating an unordered object set with an ordered object set.rnA new geneic algorithm for robot task planning in an unstructured environment characterized by a parallel evolution strategy, with the main goal of spatial transformation of an unordered object set.

• Keywords: robotika; autonomni sustavi; teorija adaptivne rezonancije; genetski algoritmi
• Keywords (english): robotics; autonomous systems; Adaptive resonance theory; genetic algorithms
• Language: croatian
• URN:NBN: urn:nbn:hr:235:161283
• Study programme: Title: Mechanical Engineering and Naval Architecture; Study programme type: university; Study level: postgraduate; Academic / professional title: doktor/doktorica znanosti, područje tehničkih znanosti (doktor/doktorica znanosti, područje tehničkih znanosti)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Open access
• Created on: 2020-04-24 07:44:16