Target Tracking Using Optical Markers for Remote Handling in ITER
Gonçalves Ribeiro, Laura (2023)
Gonçalves Ribeiro, Laura
Tampere University
2023
Tieto- ja sähkötekniikan tohtoriohjelma - Doctoral Programme in Computing and Electrical Engineering
Informaatioteknologian ja viestinnän tiedekunta - Faculty of Information Technology and Communication Sciences
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Väitöspäivä
2023-06-02
Julkaisun pysyvä osoite on
https://urn.fi/URN:ISBN:978-952-03-2882-5
https://urn.fi/URN:ISBN:978-952-03-2882-5
Tiivistelmä
The thesis focuses on the development of a vision system to be used in the remote handling systems of the International Thermonuclear Experimental Rector - ITER. It presents and discusses a realistic solution to estimate the pose of key operational targets, while taking into account the specific needs and restrictions of the application.
The contributions to the state of the art are in two main fronts: 1) the development of optical markers that can withstand the extreme conditions in the environment; 2) the development of a robust marker detection and identification framework that can be effectively applied to different use cases. The markers’ locations and labels are used in computing the pose.
In the first part of the work, a retro reflective marker made up ITER compliant materials, particularly, fused silica and stainless steel, is designed. A methodology is proposed to optimize the markers’ performance. Highly distinguishable markers are manufactured and tested.
In the second part, a hybrid pipeline is proposed that detects uncoded markers in low resolution images using classical methods and identifies them using a machine learning approach. It is demonstrated that the proposed methodology effectively generalizes to different marker constellations and can successfully detect both retro reflective markers and laser engravings.
Lastly, a methodology is developed to evaluate the end-to-end accuracy of the proposed solution using the feedback provided by an industrial robotic arm. Results are evaluated in a realistic test setup for two significantly different use cases.
Results show that marker based tracking is a viable solution for the problem at hand and can provide superior performance to the earlier stereo matching based approaches. The developed solutions could be applied to other use cases and applications.
The contributions to the state of the art are in two main fronts: 1) the development of optical markers that can withstand the extreme conditions in the environment; 2) the development of a robust marker detection and identification framework that can be effectively applied to different use cases. The markers’ locations and labels are used in computing the pose.
In the first part of the work, a retro reflective marker made up ITER compliant materials, particularly, fused silica and stainless steel, is designed. A methodology is proposed to optimize the markers’ performance. Highly distinguishable markers are manufactured and tested.
In the second part, a hybrid pipeline is proposed that detects uncoded markers in low resolution images using classical methods and identifies them using a machine learning approach. It is demonstrated that the proposed methodology effectively generalizes to different marker constellations and can successfully detect both retro reflective markers and laser engravings.
Lastly, a methodology is developed to evaluate the end-to-end accuracy of the proposed solution using the feedback provided by an industrial robotic arm. Results are evaluated in a realistic test setup for two significantly different use cases.
Results show that marker based tracking is a viable solution for the problem at hand and can provide superior performance to the earlier stereo matching based approaches. The developed solutions could be applied to other use cases and applications.
Kokoelmat
- Väitöskirjat [4946]