Sistema de navegación en interiores para asistir en la movilidad de personas con discapacidad que utilizan sillas de ruedas mediante tecnología LIFI y Machine Learning.

This research project addresses the issue of mobility for people with disabilities who use wheelchairs as a means of transportation in indoor environments. An indoor navigation system was designed and implemented for a wheelchair in a controlled setting, using machine learning algorithms. Additionally, a LiFi device was developed to assist in locating the wheelchair. The project was carried out in three phases. In the first phase, the most suitable algorithms, architectures, and devices were selected. An analysis was also conducted on the algorithms and the conditions that could affect them when integrated into the final system. In the second phase, the device was assembled using a Raspberry Pi Zero W, two boards based on the Atmega328p microcontroller, and the necessary electronics for operating the navigation system based on computer vision, convolutional neural networks, Q-Learning, and LiFi technology. In the third phase, functional tests were conducted in a controlled environment under two configurations: one with random movement and the other using machine learning. The results demonstrated that combining LiFi with machine learning improves the wheelchair's travel time from one location to another, reducing it from 242.76 seconds to 78.26 seconds compared to the LiFi and random algorithm combination.

• Главный автор: Vasco Viteri, Dagmar Mishelle (author)
• Формат: bachelorThesis
• Язык: spa
• Опубликовано: 2025
• Предметы: Redes neuronales convolucionales; Visión artificial; LiFi; Aprendizaje automático
• Online-ссылка: http://dspace.unach.edu.ec/handle/51000/15603