Banca de QUALIFICAÇÃO: NOÉLIO HELUY FERREIRA

Uma banca de QUALIFICAÇÃO de MESTRADO foi cadastrada pelo programa.
STUDENT : NOÉLIO HELUY FERREIRA
DATE: 20/09/2024
TIME: 08:30
LOCAL: Remoto via Teams
TITLE:

Using LSTM Network Models for Noise Reduction in Inertial Measurement Unit Sensors for Simulators

KEY WORDS:

Simulation, Inertial Sensors, RNN , Machine Learning, LSTM, RNN

PAGES: 78
BIG AREA: Ciências Exatas e da Terra
AREA: Ciência da Computação
SUBÁREA: Sistemas de Computação
SPECIALTY: Software Básico
SUMMARY:

This work presents a new approach for stabilizing data col lected from low-cost Inertial Measurement Unit (IMU) sensors. Currently, advances in the use of Machine Learning are significant for refining both short-term and long-term stabilization, utilizing the Long Short-Term Memory (LSTM-RNN) algorithm to correct noise added to the data caused by interfering electromagnetic fields, which lead to instability and constant drift in sensor orientation. The intention is to combine this approach with conventional noise removal filters and low-pass filters, as these methods serve as a fundamental starting point for the proposed solution to refine inertial sensor data in real-time applications. Training the model and applying it to dynamic data from the sensor could be highly relevant in low-cost projects, enhancing accuracy through embedded software technology, rather than relying on high-precision hardware investments, which would result in high project costs. The outcomes of this work could be significant for application in simulators, which use motion platforms or simulated shooting weapons, or for application in real structures or hardware, such as UAV, large structural platforms, robotics, among others in the field of industry and engineering.

COMMITTEE MEMBERS:
Presidente - 1706731 - ALETEIA PATRICIA FAVACHO DE ARAUJO VON PAUMGARTTEN
Interno - 1072256 - EDISON ISHIKAWA
Externa à Instituição - LARISSA DE MATOS GUEDES - SEEDF
Externa à Instituição - MARIA CLICIA STELLING DE CASTRO - UERJ