Banca de DEFESA: WILLIAM DA CUNHA RIBEIRO

Uma banca de DEFESA de MESTRADO foi cadastrada pelo programa.
STUDENT : WILLIAM DA CUNHA RIBEIRO
DATE: 14/06/2024
TIME: 09:00
LOCAL: Videoconferência na plataforma Google Meet: meet.google.com/yrf-wyzw-gdf
TITLE:

Autonomous Robot for Supervision Rounds of Industrial Analog Pressure Gauges: A Computer Vision Approach

KEY WORDS:

Autonomous Robot, Manometer Identification, Odometry, Visual Landmarks, Navigation in Industrial Environments, Image Processing.

PAGES: 76
BIG AREA: Engenharias
AREA: Engenharia Elétrica
SUBÁREA: Eletrônica Industrial, Sistemas e Controles Eletrônicos
SPECIALTY: Automação Eletrônica de Processos Elétricos e Industriais
SUMMARY:

This study aims to develop a robotic system specialized in autonomously carrying out rounds to monitor the status of analog pressure gauges in industrial environments, using computer vision techniques to capture data from these devices. The adoption of this approach is justified by the complexity and cost associated with replacing analog pressure gauges with devices connected to the automation network. This replacement would imply production stoppages, resulting in lost profits for the industry during the retrofit process. Furthermore, the advantage stands out in relation to rounds traditionally carried out by human operators, which consume time and interfere with more strategic intellectual activities, such as production data analysis, performance control and preventive activities in the factory. The robot's localization is based on the fusion of odometry and localization through landmarks, using the ArUco tag standard. This approach provides a visual reference for the robot's autonomous navigation, allowing the calibration of odometry and, consequently, the correction of its location in the work environment. The embedded hardware and software architecture is crucial for the efficient functioning of the system. Odometry and control of the robot's motors are performed on a dedicated microcontroller (Arduino), while high-processing tasks such as computer vision, landmark localization and navigation are performed on an embedded computer (Raspberry Pi). This distribution of tasks optimizes system performance, allowing them to work in parallel. Recognizing the pressure gauges and determining the measurement from captured images are fundamental to the robot's purpose. This process involves recognizing the circumference of the pressure gauge for identification, followed by capturing the image for subsequent processing. This approach ensures efficient data collection without burdening the robot at the time of capture. This work presents an innovative solution for the supervision of analog pressure gauges in industrial environments, highlighting the effectiveness of computer vision and robotic autonomy. The proposed approach not only optimizes financial resources, avoiding costly pressure gauge replacement, but also increases operational efficiency, freeing operators for more strategic tasks.

COMMITTEE MEMBERS:
Presidente - 1242315 - PABLO JAVIER ALSINA
Interno - 4523538 - MARCELO BORGES NOGUEIRA
Interno - 2718169 - MARCIO VALERIO DE ARAUJO
Externo ao Programa - 1241170 - HEITOR MEDEIROS FLORENCIO - UFRNExterno à Instituição - ANFRANSERAI MORAIS DIAS - UEFS