The MTMCA Research Group focuses on methods and techniques to approximate science and engineering when handling modern cyber-physical automation systems (Fig. 1). That is, we intend to integrate more harmonically and efficiently the signal level, and the information processing level, considering the multiple challenges that this integration imposes, e.g., control, communication, quality of signals and events, refinement of events, information fusion and decomposition, cognition, intelligence.
Fig. 1: Usual Automation & control architecture.
Essentially, we aim to develop and apply computational methods and tools to solve real life engineering problems. Our application domains include distributed power generation systems, flexible manufacturing, and precision agriculture.
This video-guide is part of the review process for the paper:
Flexible Control of Discrete Event Systems using Environment Simulation and Reinforcement Learning
submitted to the Applied Soft Computing journal, under review.
This project aims to combine the safety nature of the Supervisory Control Theory (SCT) of Discrete Event Systems (DESs), with flexibility and smart decisions returned from machine learning approaches.
Technically, we consider to split the set of events of a DES into two sets: probabilistic and non-probabilistic events. Then, the first set is handled using machine learning, while the second set is approached with SCT. After information processing, the two branches are merged and combined to provide safe, but smart, decisions to the target DES.
Implementation, communication and cyber-physical aspects are also topics that we aim to consider in future researches.
This project aims to combine safety properties from the Supervisory Control Theory (SCT) of Discrete Event Systems (DESs) with flexible rules and planning estimated using stochastic models.
Implementation, communication, and cyber-physical aspects are also topics that we aim to consider in future research.
A poultry condominium consists of poultry houses that are constructed in proximity and share an isolated, controlled, modular, and escalable infrastructure.
Each poultry house belonging to the condominium is managed by a different specialist, which implies divergence in the action plan applied by each one.
Our research area is focused on cooperation among different poultry houses, in an attempt for optimizing their action plans. For this purpose, the focus of our research is:
(i) the integration of different learning methods;
(ii) the application of this integration on poultry management; and
(iii) the implementation of a practical control solution that includes the control, communication and supervision parts.
Technical Cooperation Project, between UTFPR and the Municipality of Vacaria - RS.
The main objective is to use Data Mining for municipal public management, coming from the Municipality's databases in the most diverse areas of activity, aiming at obtaining a tool for management and strategy in decision making in the public area.
The project aims to evaluate, propose, develop and implement an intelligent data crossing system to assist the public management of the Municipality of Vacaria through computational intelligence techniques in order to develop a support system for managerial decisions.
The system receives a set of inputs of real data, coming from the databases provided by the municipality, and return as output indicators that favor the decision-making process by the Municipal Administration.
The Araucária & Renault do Brasil Scholarship Program encourages the articulation among higher educational institutions and Renault do Brasil company, providing a partnership in the training of future professionals; as well as promoting student learning in a practical and real environment related to the automobile universe.
The MTMCA currently has three active researchers working at the Renault plant in São José dos Pinhais, Parana, Brazil.
One of the projects is developed by Cristian R. Pastro and seeks solutions for the extraction and treatment of data from industrial processes in automotive painting.
The project developed by student Joceleide D.C. Mumbelli seeks to improve a Computer Vision System existing on the production line. The existing computer vision system is provided by the company Keyence and it has shown compliance problems. The focus of the work is to use Deep Learning techniques to improve the results obtained by online production equipment.
Henrique Ogata works on a project related to Industrie 4.0. The objective is to seek new technologies based on LoRa, RFID and digitalization of processes to be implemented in the company.
The objective here is to use didactic-scale production cells for exploiting properties commonly found in manufacturing processes, such as concurrency, synchronous and asynchronous production, priority policies, justice, etc. by properly treating these properties, from the perspective of control and automation resources, we try to increase productivity to manufacturing processes, while robustness and safety are preserved.
This project starts from modelling issues, it evolves through different synthesis approaches, reaches code generation and hardware programming domains, and ends up at the supervision systems and Information Technology support.