Digital twins: uma célula de manufatura inteligente na indústria 4.0

• Ano de defesa: 2025
• Autor(a) principal: Perez, Leonardo Mazetto
• Orientador(a): Gatti, Daniel Couto
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Pontifícia Universidade Católica de São Paulo
• Programa de Pós-Graduação: Programa de Pós-Graduação em Tecnologias da Inteligência e Design Digital
• Departamento: Faculdade de Ciências Exatas e Tecnologia
• País: Brasil
• Palavras-chave em Português: Simulação digital; Manufatura inteligente; Visão computacional; Automação; Ciberfísico
• Palavras-chave em Inglês: Digital simulation; Intelligent manufacturing; Computer vision; Automation; Cyber-physica
• Área do conhecimento CNPq: CNPQ::ENGENHARIAS
• Link de acesso: https://repositorio.pucsp.br/jspui/handle/handle/45401

The advancement of digital technologies within the context of Industry 4.0 has enabled new forms of integration between physical and virtual systems. However, their practical implementation still faces limitations related to cost, complexity, and accessibility in experimental environments. This study is based on the hypothesis that the implementation of emerging technologies—such as digital twins, computer vision, and embedded control—within a compact physical cell can enhance predictability, control, and understanding of intelligent manufacturing processes. The objective of this dissertation is to develop and validate a small-scale intelligent manufacturing cell, designed independently and built with open-source technologies and low-cost components. The research adopts a technical and experimental approach, aiming to demonstrate the feasibility of integrating cyber-physical systems, sensors, actuators, and real-time digital simulation. Although the proposed model shows potential for academic use, its primary focus is not educational, but rather the practical demonstration and functional validation of the integration among key technological pillars of Industry 4.0 within a controlled laboratory environment. Methodologically, this is an applied and exploratory study combining qualitative and quantitative approaches. The research involved three-dimensional modeling, physical prototyping, the development of a convolutional neural network (YOLO) for object detection, and the construction of a digital twin synchronized in real time with the physical system. The results demonstrate that the developed model is functional, modular, and replicable, operating in an integrated manner between sensors, actuators, and digital simulation. The findings indicate that the proposed system can serve as a foundation for future applications in robotic autonomy, predictive maintenance, and integration with reference architectures of Industry 4.0