Controle e localização de robô móvel via aplicativo Android: uso de OpenCV e KNN
Data
Autores
Título da Revista
ISSN da Revista
Título de Volume
Resumo
Este trabalho apresenta o desenvolvimento de um aplicativo móvel para a plataforma Android destinado ao controle manual de um robô e ao reconhecimento de ambientes internos por meio de técnicas de visão computacional. O aplicativo permite a comunicação com o robô por comandos de rede, o controle de suas funções principais via socket TCP e a exibição do streaming de vídeo no smartphone via requisição HTTP. Também foi implementado um subsistema de reconhecimento de ambientes utilizando a biblioteca OpenCV em conjunto com os algoritmos LBP e KNN. Nesse subsistema, os ambientes são previamente mapeados a partir de quatro imagens correspondentes às direções norte, leste, sul e oeste, sendo posteriormente comparados com novas imagens para estimar o ambiente atual. Na versão validada neste trabalho, as imagens utilizadas no reconhecimento foram obtidas a partir da galeria do smartphone, embora a implementação tenha sido organizada para permitir, futuramente, a substituição dessa fonte por frames do vídeo do robô. Os resultados indicaram que a solução foi funcional no escopo avaliado, demonstrando potencial para aplicação de uma abordagem de baixo custo no reconhecimento de ambientes internos.
This work presents the development of a mobile application for the Android platform designed for the manual control of a robot and the recognition of indoor environments using computer vision techniques. The application allows communication with the robot via network commands, control of its main functions via TCP socket, and display of streaming video on the smartphone via HTTP request. An environment recognition subsystem was also implemented using the OpenCV library together with the LBP and KNN algorithms. In this subsystem, environments are previously mapped from four images corresponding to the north, east, south, and west directions, and subsequently compared with new images to estimate the current environment. In the version validated in this work, the images used for recognition were obtained from the smartphone's gallery, although the implementation was organized to allow, in the future, the replacement of this source with frames from the robot's video. The results indicated that the solution was functional within the evaluated scope, demonstrating potential for applying a low-cost approach to indoor environment recognition.
This work presents the development of a mobile application for the Android platform designed for the manual control of a robot and the recognition of indoor environments using computer vision techniques. The application allows communication with the robot via network commands, control of its main functions via TCP socket, and display of streaming video on the smartphone via HTTP request. An environment recognition subsystem was also implemented using the OpenCV library together with the LBP and KNN algorithms. In this subsystem, environments are previously mapped from four images corresponding to the north, east, south, and west directions, and subsequently compared with new images to estimate the current environment. In the version validated in this work, the images used for recognition were obtained from the smartphone's gallery, although the implementation was organized to allow, in the future, the replacement of this source with frames from the robot's video. The results indicated that the solution was functional within the evaluated scope, demonstrating potential for applying a low-cost approach to indoor environment recognition.
Descrição
Palavras-chave
Citação
SILVA, Lucas da. Controle e localização de robô móvel via aplicativo Android: uso de OpenCV e KNN. 2026. Trabalho de Conclusão de Curso (Bacharelado em Engenharia de Telecomunicações) - Instituto Federal de Educação, Ciência e Tecnologia de Santa Catarina, São José, 2026.
