Complete coverage of large seabed areas using sonar data-based path planning for an autonomous marine vehicle

doctoral thesis

• Title (english): Planiranje putanje autonomnoga plovila zasnivano na sonarskim podacima u svrhu potpunoga prekrivanja velikih površina morskoga dna
• Author: Nadir Kapetanović
• Mentor: Nikola Mišković (mentor)
• Mentor: Adnan Tahirović (komentor); VIAF: 7119164479567526210007
• Committee member: Zdenko Kovačić (predsjednik povjerenstva)
• Committee member: Đula Nađ (član povjerenstva)
• Committee member: Fabio Bruno https://orcid.org/0000-0002-9755-475X (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; (Department of Control and Computer Engineering); Zagreb
• Defense date and country: 2023-06-26, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Electrical Engineering
• Universal decimal classification (UDC): 621.3 - Electrical engineering

Abstract

Efficient mapping of an unknown large-scale marine area using a side-scan sonar onboard an autonomous marine vehicle is often of great importance. It might also be important to scan parts of an area in more detail and from more than one side. In contrast to the standard offline static coverage problem solution for side-scan sonar missions based on the overlapping all across-track sonar swaths in a lawnmower pattern, several online sonar-data driven coverage path planning algorithms are proposed in this thesis. The proposed algorithms provide a coverage solution based on local information gain computed from the side-scan sonar data during the mission execution, which is then exploited for replanning. In addition, the proposed solution takes into account the coverage path length/time to ensure equal information content about arbitrarily defined interesting objects as in case of the standard lawnmower with much less resources needed to complete the same mission.

One dynamical programming-based and three decision making-based coverage path (re)planning algorithms are proposed herein. Their performance was thoroughly examined in a grid-like coverage area cost map simulation over a whole range of coverage mission parameters and randomly generated seafloor configurations to gain statistical metrics of their behavior. Also, their upper and lower performance bounds are mathematically modelled w.r.t. some mission parameters, and were validated by the thousands of simulation runs. Moreover, an existing realistic 3D simulation environment was extended by the above-mentioned coverage path planners, mission controller, and sonar data processing modules to gain further insights into the interplay of coverage path planning, feasible path interpolation, vehicle dynamics and control.

Abstract (croatian)

Učinkovito mapiranje velikih nepoznatih područja morskoga dna korištenjem bočno skenirajućeg sonara (engl. side-scan sonar) i autonomnoga pomorskog vozila je često od velikog značaja. Također, često je bitno detaljnije snimiti zanimljive dijelove tog područja i to s više od jedne strane. Predloženo je nekoliko online metoda prekrivanja područja zasnovanih na sonarskim podacima. Razvijene metode su poređene sa standardnim offline rješenjem problema prekrivanja sonarom, tzv. uzorkom kosilice koji sve objekte neadaptivno snima sa obje strane bez obzira da li uopće postoje ikakvi objekti na morskome dnu, i ako postoje da li su zanimljivi za trenutnu misiju. Za razliku od toga, predloženi algoritmi nude rješenje problema (re)planiranja prekrivanja područja zasnovano na sonarskim podacima u tijeku misije. Nadalje, oni u dosta velikom području vrijednosti parametara misije nude jednako informativnu snimku interesantnih objekata kao i neadaptivni pristup uz znatno skraćenje trajanja misije/dužine putanje prekrivanja.

Jedan algoritam zasnovan na dinamičkom programiranju i tri algoritma zasnovana na heurističkom donošenju odluka su predloženi u ovom radu. Njihove performanse su detaljno ispitane na matričnoj mapi cijena prekrivanja i to za širok opseg vrijednosti parametara misije i slučajno generarirane konfiguracije morskoga dna kako bi se dobio uvid u statistiku ponašanja navedenih algoritama. Također, njihove gornje i donje granice performansi su matematički opisane u ovisnosti od parametara misije, te su validirane tisućama provedenih simulacija. Štoviše, postojeće realistično 3D simulacijsko okruženje je prošireno sa gore navedenim planerima prekrivanja, kontrolerom misije, i modulom obrade sonarske slike kako bi se stekao još i bolji uvid u međusobne utjecaje modula planiranja putanje prekrivanja, interpolacije izvedivih putanja, te dinamike i upravljanja vozilom.

• Keywords: autonomous underwater vehicles; coverage path planning; sensor data-driven coverage; side-scan sonar; image processing; anomaly detection; target detection; saliency
• Keywords (croatian): autonomna podvodna vozila; planiranje putanje prekrivanja; prekrivanje bazirano na senzorskim podacima; bočno-skenirajući sonar; procesiranje sonarskih slika; detekcija anomalija; detekcija meta; detekcija istaknutih objekata; modelski prediktivno upravljanje; podvodna lokalizacija
• Language: english
• URN:NBN: urn:nbn:hr:168:941026
• Promotion: 2023
• Study programme: Title: Doctoral study programme "Electrical Engineering and Computing"; Study programme type: university; Study level: postgraduate; Academic / professional title: doktor/doktorica znanosti, po-dručje tehničkih znanosti (doktor/doktorica znanosti, po-dručje tehničkih znanosti)
• Type of resource: Text
• Extent: xii, 196 str. : graf. prikazi ; 31 cm.
• File origin: Born digital
• Access conditions: Open access
• Created on: 2024-02-08 11:52:38