An Approach for Modeling and Simulation of Virtual Sensors in Automatic Control Systems Using Game Engines and Machine Learning

Sensors (Basel). 2024 Nov 28;24(23):7610. doi: 10.3390/s24237610.

Abstract

We live in an era characterized by Society 4.0 and Industry 4.0 where successive innovations that are more or less disruptive are occurring. Within this context, the modeling and simulation of dynamic supervisory and control systems require dealing with more sophistication and complexity, with effects in terms of development errors and higher costs. One of the most difficult aspects of simulating these systems is the handling of vision sensors. The current tools provide these sensors but in a specific and limited way. This paper describes a six-step approach to sensor virtualization. For testing the approach, a simulation platform based on game engines was developed. As contributions, the platform can simulate dynamic systems, including industrial processes with vision sensors. Furthermore, the proposed virtualization approach allows for the modeling of sensors in a systematic way, reducing the complexity and effort required to simulate this type of system.

Keywords: automatic control systems; game engines; industry 4.0; machine learning; systems modeling and simulation; systems virtualization.

Grants and funding

This research was funded by the UNINOVA research institute (https://www.uninova.pt, accessed on 15 November 2024, within the Research unit CTS-Centro de Tecnologia e SistemasUIDB/00066/2020).