The best research projects of students have been selected | Gdańsk University of Technology

As part of the first edition of research projects, which is a continuation of the subject Group project in the second cycle studies at FETI, over 70 projects have been successfully implemented. The jury of the competition for the "best project at FETI" awarded three prizes and four dean's distinctions, and IHS Markit company awarded "the best project in the field of artificial intelligence". The award ceremony and presentation of the projects took place during the inauguration of the second-cycle studies at the faculty. The invited students were welcomed by prof. Jacek Stefański, dean of the FETI.

– We are on the threshold of the 70th anniversary of our faculty. You start your studies in the jubilee year of our faculty and I am very happy to welcome a really large group of students. You face studies that last only three semesters. It is a short, but extremely important time for you and for your future professional career, which opens the door to the labor market.

After the official opening, cash prizes and gifts were presented to the best student research teams.

– The authors of the projects showed not only innovation, but also diligence. What's more, several papers have been published in renowned journals or at global conferences, and another dozen or so papers have a chance for publication (some are already under review) - said Krzysztof Nowicki, PhD, Eng., professor at Gdańsk Tech, plenipotentiary of the Dean of FETI for research projects.
 

Awards

First place

• Implementation of the prototype of the WBAN radiolocation system with the use of deep learning

manager: Sebastian Urwan

team members: Dominika Wysocka, Alicja Pietrzak

supervisor: Krzysztof Cwalina PhD, Eng.

The team made a prototype of a hybrid radiolocation system for people, using deep learning methods, which is a response to the needs of the SALUBRE Rehabilitation and Care Center in Rumia and was developed in accordance with the company's requirements and guidelines. The effectiveness of the developed system was verified on the basis of measurements in real indoor and outdoor propagation environments, and the obtained results (as well as the system itself) are innovative and are the subject of a prepared publication in an international journal from the ministerial list.

Second place

• Application of machine learning algorithms to detect cancer in liquid biopsy

manager: Franciszek Górski

team members: Piotr Juszczyk, Sebastian Lewalski

supervisor: Sebastian Cygert MSc, Eng.

The aim of the project was to create an algorithm for the detection of cancer in a liquid biopsy. The project was carried out in cooperation with the Medical University of Gdańsk and was aimed at verifying the innovative approach used by the MUG team to early cancer detection. The innovative approach was to use liquid biopsy data to train the neural network. The challenge and, at the same time, the innovation of the project was the new application of machine learning algorithms. Working with real medical data and their limited availability was an additional challenge. The effect of the project is, among others the work presented during the 4th workshop on “PRedictive Intelligence in MEdicine” (PRIME) at MICCAI 2021.

Third place

• Supporting the safety of people and car intelligence with the use of automatic pedestrian detection in thermal image sequences

manager: Aleksandra Górska

team members: Iga Namiotko, Martyna Włoszczyńska, Aleksandra Górska, Aleksandra Wędołowska, Patrycja Guzal

supervisor: Jacek Rumiński PhD, DSc, Eng., professor at Gdańsk Tech

Detecting pedestrians, especially in difficult road conditions, is an important research goal in the development of autonomous vehicles. This task was undertaken by students of the artificial intelligence specialization while designing the thermal imaging module. By placing the module on the car, they captured thousands of images that they described and used in numerous adapted machine learning models. As a result, they developed a complete prototype of an innovative solution enabling effective detection of pedestrians even at night. The result of the project are, among others, two works submitted to the IEEE / IES Human-System Interaction conference, Australia 2022.

Distinctions

• Rapid automatic design of planar antennas using optimization algorithms

manager: Michał Czyż

team members: Kacper Błaszkowski

supervisor: Adrian Bekasiewicz PhD, DSc, Eng., professor at Gdańsk Tech

The aim of the project was to develop solutions enabling the automatic generation of planar antenna topologies that meet specific requirements for operating parameters. The task was carried out by developing an innovative topology of a universal planar antenna, the geometry of which is represented using a combination of trigonometric functions, thus enabling the generation of various geometries while ensuring a limited number of input parameters. The developed universal model of the antenna was used in conjunction with an innovative classifier and efficient optimization tools using substitute models, thus allowing for the automatic building of a set of radiators while maintaining a low cost of the procedure. Two papers will be presented at the end of March at an international conference related to antenna engineering (16th European Conference on Antennas and Propagation, Madrid 2022). An article in the A-list of the Ministry of Science and Higher Education “International Journal of Electronics and Telecommunications” has also been accepted for publication.

• Indoor location system supporting the movement of blind people

manager: Michał Wysocki

team members: Marta Trzaska, Robert Nicpoń

supervisor: Agnieszka Czapiewska PhD, Eng.

The aim of the project was to test the usefulness of an indoor location system using Bluetooth Low Energy (BLE) technology as a support for blind people inside or near large buildings. As part of the project, measurement tools for the fingerprinting method were implemented, a number of measurements in real conditions and an analysis of the positioning accuracy of the selected method was carried out. The innovativeness of the research carried out is related to the implementation of measurements for various scenarios of device configuration and the use of a large number of different terminals.

• A small surface unmanned vessel (mUSV) capable of independently completing the planned route

manager: Cezary Sobczak

team members: Michał Niemc, Mikołaj Dyrka, Patryk Monarcha, Marek Kasiewicz

supervisor: Henryk Lasota PhD, DSc, Eng., professor at Gdańsk Tech

For the needs of the laboratory and design studio of a new specialization supplementing Autonomous Vehicle Technologies (AVT), the team constructed (as part of the Department of Sonar Systems, in cooperation with Navinord) a floating platform mUSV (mini Unmanned Surface Vehicle). It is a universal carrier of built-in devices and systems - navigation and observation ones, operating in real time, enabling operation in an autonomous mode, which means advanced automation enriched with the ability to recognize obstacles, modify the planned route and, after avoiding an obstacle, continue the assumed mission. During operation, the platform reports its parameters via a radio link (position, direction, speed, battery charge level, error information).

• Investigation of entropy properties in entangled systems

manager: Jan Horodecki

supervisor: Piotr Mironowicz PhD, Eng.

The subject of the research was to determine the relationship between two phenomena characteristic for quantum systems: breaking Bell's inequality and negative conditional entropy. These issues concern the very essence of quantum information processing and quantum communication. As part of the research project, Eng. Jan Horodecki showed that there is an interesting relationship between the directly observable value, such as the violation of Bell's inequality, and the directly unobservable value, which is the negativity of conditional entropy. This discovery may be the basis for the development of new communication and information processing protocols in the future, hence it may potentially be quite important for ICT. The work is expected to be submitted to Quantum Information Processing within the next few weeks.

Awards for projects in the field of AI

Two prizes sponsored by IHS Markit company were awarded in the competition for "Best Artificial Intelligence Design".

First prize

• Implementation of the prototype of the WBAN radiolocation system with the use of deep learning

This project received a total of the first two awards, see the description above.
 

Second prize

• Analysis of infrared images and segmentation of facial features on thermograms using AI for the purposes of COVID-19 prevention

manager: Piotr Filipowicz

team members: Kinga Słomińska, Piotr Filipowicz, Mateusz Kowalewski

supervisor: Mariusz Kaczmarek PhD, DSc, Eng., professor at Gdańsk Tech

The aim of the project was to conduct an analysis of the state of knowledge regarding the use of machine learning methods to detect the face area and characteristic points of the face (eyes) in images, in particular thermographic images (infrared imaging), and then the implementation of the face recognition module for the application of the temperature screening monitoring system in people entering the room. This is a complex task because the software module is to be used to automatically designate the face area and characteristic elements of the face on the thermograms, taking into account the distance of the person from the thermographic camera. Additionally, it is assumed that people enter the room from outside, where the weather conditions change (depending on the time of day, season and aura), which results in different temperature distributions on the face. The results of the work were submitted to the journal from the ministerial list "IEEE Transactions on Biomedical Engineering"

The projects were carried out as part of the subject Research project in second-cycle studies at FETI, but students from other faculties took part in it as well. The aim of the course is to prepare students for future work in a research team and to teach the timely fulfillment of obligations arising from the schedule. Project topics are formulated by any clients, including those from outside Gdańsk Tech, and these may be application projects as well as application and research projects. Students verify the research hypothesis set by the client, which may require the implementation of a product, e.g. an application, device, and carrying out appropriate research and analysis of the results. Then, the authors of the project submit the report in the form of a publication in English or - in the case of application projects - in the form of a patent application.