Uma interface com foco em usabilidade para a relação entre homem e inteligência artificial para um sistema de automação industrial: desenvolvimento de uma solução tecnológica com foco na usabilidade para a relação entre usuário e inteligência artificial para um sistema de automação industrial

Abstract

The search for the optimization of production processes has always been present in the industry’s history. The consequences of this incessant search have brought numerous innovations in the scientific and technological fields. Throughout time, this search for bigger and better results has never diminished, on the contrary, it has always been greater with each industrial revolution. Today, in Industry 4.0, the presence of advanced technologies coming from the scientific universe has potentiated even more the gains for the industrial branch.But finding a middle ground between greater technology and ease of use is one of the challenges when it comes to modern, high-performance machines. One of the scenarios found in the industrial context is the interaction between man and machine, which is increasingly present. So you can see that this partnership needs to be consistent and easy to understand.Particularly when it comes to using computer vision algorithms to recognize objects in images, this requires some prior knowledge in computer science and/or programming languages. With this, the present proposes a user-friendly interface for a system that should facilitate the training of a convolutional a convolutional neural network in the detection of crystals in microscopic images of clinker, which is a product used in the production of Portland cement. Thus, with the use of the proposed proposed tool, a person without much skills in programming language will be able to train and use an algorithm for object detection, without great difficulties and in an intuitive way. For this, methodologies and tools that facilitate the development of the proposed the development of the proposed system, besides potentiating the expected results. With this, the present work aims to show that it is possible that people with little or no computational knowledge are able to use an on-line platform destined to the use of these algorithms. That is, this person trains and runs an Artificial Intelligence, without great difficulty and in an intuitive way. To this end, methodologies and tools were studied to facilitate the development of the proposed system, and to enhance the expected results.