Sustav za automatsku detekciju oštećenja na infrastrukturi

master's thesis

• Title (english): System for automatic detection of infrastructure damage
• Author: Andro Katanec
• Mentor: Matko Orsag (mentor)
• Committee member: Matko Orsag (član povjerenstva)
• Committee member: Frano Petric (član povjerenstva)
• Committee member: Tamara Petrović (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2023-07-14, Croatia

Abstract

U radu je implementiran sustav za automatsku detekciju oštećenja na infrastrukturi pomoću bespilotne letjelice opremljene dubinskom kamerom. Detekcija oštećenja realizirana je pomoću YOLOv7 modela koji je treniran na skupu podataka od ukupno 11.298 slika pukotina. Za implementaciju algoritama za automatsku detekciju korišteno je ROS okruženje povezano s Gazebo simulatorom unutar kojeg je postavljen model mosta s pločama na kojima se nalaze pukotine. Implementirani su ROS čvorovi za detekciju pukotina, procesiranje i vizualizaciju pukotina te za autonomno kretanje letjelice. Testiranje sustava provedeno je unutar Gazebo simulatora na modelu osječkog pješačkog mosta te su prikazani ostvareni rezultati.

Abstract (english)

This thesis implements a system for automatic detection of infrastructure damage using a UAV equipped with a depth camera. Damage detection is achieved using the YOLOv7 model trained on a dataset of 11.298 images containing cracks. A ROS environment connected to a Gazebo simulator was utilized for implementing the algorithms for automatic detection. A bridge model with tiles containing cracks was set up within the simulator. ROS nodes were implemented for crack detection, processing and visualization of detected cracks, as well as for autonomous navigation of the UAV. The system was tested inside the Gazebo simulator using a model of an Osijek pedestrian bridge and the achieved results are presented.

• Keywords: YOLOv7; detekcija oštećenja; ROS; bespilotna letjelica; Gazebo
• Keywords (english): YOLOv7; damage detection; ROS; unmanned aerial vehicle; Gazebo
• Language: croatian
• URN:NBN: urn:nbn:hr:168:101456
• Study programme: Title: Information and Communication Technology; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka informacijske i komunikacijske tehnologije (magistar/magistra inženjer/inženjerka informacijske i komunikacijske tehnologije)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-10-11 09:16:43