Bildebehandling og Autonomi

Abstract

Abstrakt. Denne bachelor oppgaven omhandler utvikling og implementering av programvare for en undervanns robot (ROV), som tar i bruk blide behandling og data syn for å utføre autonome oppgaver. Dette gjøres som en en del av et tverrfaglig prosjekt i student organisasjonen UiS Subsea. ROV-en er bygd med mål om å delta i MATE ROV World Championship, som holdes i Colorado, USA, den 20-24. juni 2023.

Denne bachelor oppgaven beskriver hvordan kamera-strømmer mottas på topside systemet, fra ROV-en, hvordan bilde behandling blir brukt til å løse autonome oppgaver, hvordan programmet implementeres i et brukergrensesnitt (GUI) og hvordan kjøre kommandoer sendes ned til ROV-en.

Et modulært system har blitt laget, som mottar flere kamera strømmer, utfører autonome oppgaver og sender styre kommandoer. Programmet er blitt implementert inn i en GUI og testet på land. Tydelige definerte bildebehandlings oppgaver er gitt ut av MATE. Alle oppgavene er løst, utenom 3D modellering. Programmet er testet på land og oppfører seg som ønsket. Derimot, er ikke programmene testet i vann. Grunnen til dette er at ROV-en ikke var ferdigstilt i tide for vår gruppe å teste programmene våre.

Testing og forbedring av programmet vill fortsette i forberedelser til MATE. Grunnet leveringsfristen for bachelor oppgaven, blir ikke dette inkludert i denne rapporten.

Se vår GitHub repository for koden: https://github.com/UiS-Subsea/Bachelor_Bildebehandling Se GUI gruppens GitHub repository for implementeringen av vår kode inn i deres kode: https://github.com/UiS-Subsea/Bachelor_GUI

Abstract. This bachelor’s thesis is about creating and implementing a software program on an underwater robot (ROV), that utilizes image processing and computer vision to perform autonomous tasks. This is done as part of an interdisciplinary project in the student organization UiS Subsea. The ROV is built with the purpose of competing in the MATE ROV World Championship, which is held in Colorado, USA, on the 20-24th of June 2023.

This thesis describes how camera feeds are received on the topside system, from the ROV, how image processing is utilized to solve autonomous tasks, the implementation of the program into a graphical user interface (GUI), and the sending of driving commands down to the ROV. Additionally, an attempt was made to use computer vision to create a 3D model of a coral head.

A modular program has been made, receiving multiple camera feeds, performing autonomous tasks, and sending driving commands. This program was implemented into a GUI, and tested on land. Clearly defined image-processing tasks were given by MATE. All of them were solved except for 3D modeling. The program was tested on land and behaved as intended. However, the programs were not tested in water. This was due to the ROV not being finished in time for our group to test our programs.

Testing and improving the program will continue in preparation for MATE. Due to the submission date for the bachelor’s thesis, this will not be included in this paper.

See our GitHub repository for the code: https://github.com/UiS-Subsea/Bachelor_Bildebehandling See the GUI group’s repository for our code implemented into their code: https://github.com/UiS-Subsea/Bachelor_GUI