Uporaba tehnologij sledenja pogleda v zdravstvu : diplomsko delo

Pretnar, Filip (avtor), Lebar, Andrej (mentor), Debeljak, Mojca (komentor), Lampe, Tomaž (recenzent)

Uvod: Osebe z zmanjšano ali odsotno mobilnostjo zgornjih udov imajo lahko težave pri samostojnem opravljanju vsakodnevnih dejavnosti. Podporne robotske roke lahko del teh omejitev zmanjšajo, vendar njihovo upravljanje pogosto zahteva ohranjeno mobilnost drugih delov telesa. Sledenje pogledu predstavlja možnost brezkontaktnega upravljanja, pri katerem uporabnik ukaze izbira s pogledom. Namen: Namen diplomskega dela je bil izdelati program s katerim lahko uporabnik upravlja robota s pogledom in preveriti učinkovitost in uspešnost delovanja tega programa. Metode dela: Izdelali smo dva programa v programskem jeziku Python. Prvi program je s knjižnico EyeGestures in spletno kamero določal položaj uporabnikovega pogleda na zaslonu. Zaslon je bil razdeljen na štiri kontrolna območja, ki so predstavljala ukaze za premikanje robotske roke. Drugi program je sprejemal številke izbranih območij in jih pretvarjal v premike robota. Programa sta delovala v ločenih virtualnih okoljih in med seboj komunicirala prek lokalne omrežne povezave. Delovanje sledenja pogledu smo preverili v 5 meritvenih sejah, v katerih je bilo izvedenih 100 meritev z očali in 100 meritev brez očal. Rezultati: Programa sta omogočala premikanje robotske roke glede na opazovano kontrolno območje. Povprečni odzivni čas je znašal 0,88 sekunde z očali in 0,96 sekunde brez očal. Razlika med pogojema je bila majhna. Med preizkušanjem so se pojavljale težave pri kalibraciji in zmanjšanje natančnosti ob premikanju glave. Razprava in zaključek: Izdelani prototip je pokazal, da je mogoče cenovno dostopno spletno kamero in programsko sledenje pogledu uporabiti za osnovno upravljanje robotske roke. Sistem je bil dovolj učinkovit za predstavitev koncepta, vendar zaradi omejene natančnosti, občasnih težav pri kalibraciji in nezanesljivosti še ni primeren za neposredno uporabo pri pacientih. Za varnejšo praktično uporabo bi bile potrebne izboljšave sledenja pogledu, povratne informacije uporabniku in dodatni varnostni mehanizmi.

• Jezik: Slovenski jezik
• Ključne besede: diplomska dela, ortotika in protetika, sledenje pogledu, podporna robotika, robotska roka, upravljanje s pogledom, Dobot Magician
• Vrsta gradiva: Diplomsko delo/naloga
• Tipologija: 2.11 - Diplomsko delo
• Organizacija: ZF - Zdravstvena fakulteta
• Kraj izida: Ljubljana
• Založnik: [F. Pretnar]
• Leto izida: 2026
• Št. strani: IX, 48 str., [13] str. pril.
• PID: 20.500.12556/RUL-186264
• UDK: 617.3
• COBISS.SI-ID: 289408003
• Datum objave v RUL: 29.08.2026

Sekundarni jezik

• Jezik: Angleški jezik
• Naslov: Use of eye tracking technology in healthcare : diploma work

Izvleček

Introduction: People with reduced or absent upper-limb mobility may experience difficulties performing activities of daily living independently. Assistive robotic arms can reduce some of these limitations; however, their operation often requires preserved mobility in other parts of the body. Gaze tracking provides a contactless control method in which users select commands with their gaze. Purpose: The purpose of this thesis was to develop a program that enables users to control a robot with their gaze and to evaluate the effectiveness and performance of the developed program. Methods: Two programs were developed in the Python programming language. The first program used the EyeGestures library and a webcam to determine the position of the user’s gaze on the screen. The screen was divided into four control zones representing commands for moving the robotic arm. The second program received the numbers of the selected zones and converted them into robot movements. The programs operated in separate virtual environments and communicated through a local network connection. The performance of the gaze-tracking system was evaluated in five measurement sessions, during which 100 measurements were performed with glasses and 100 without glasses. Results: The programs enabled the robotic arm to move according to the control zone at which the user was looking. The mean response time was 0.87 seconds with glasses and 0.96 seconds without glasses. The difference between the two conditions was small. During testing, calibration difficulties and reduced accuracy associated with head movement were observed. Discussion and conclusion: The developed prototype demonstrated that an affordable webcam and software-based gaze tracking can be used for basic control of a robotic arm. The system was sufficiently effective to demonstrate the concept; however, because of its limited accuracy, occasional calibration difficulties, and unreliability, it is not yet suitable for direct use with patients. Improvements in gaze tracking, user feedback, and additional safety mechanisms would be required for safer practical application.

• Ključne besede: diploma theses, orthotics and prosthetics, gaze tracking, assistive robotics, robotic arm, gaze-based control, Dobot Magician