Presentation title

Intelligent Cyber Physical Systems for Smart Industries: activity of the CINI LN-ESSM @ University of Messina

Authors

Dario Bruneo, Fabrizio De Vita and Antonio Puliafito

Institution(s)

Universita' di Messina - Dip. di Ingegneria, Italy

Presentation type

Presentation of a research group from one or more scientific institutions

Abstract

The advancement of the technology and the increase of the computational power have brought to the diffusion and application in the recent years of embedded systems in several scenarios. Nowadays, we live in a cyber physical world where objects interact with each other and with the environment that surrounds them. In such a context, machine learning is an interesting technique that can be applied on these devices scattered inside the environment providing them an intelligence, which allows to take decisions in an autonomous way supporting the human being. One of the main applications involving intelligent cyber physical systems is related to the industrial sector, with the aim to improve productivity and maintenance costs. In this context, the research activity of the University of Messina node of the CINI National Laboratory on Embedded Systems and Smart Manufacturing (LN-ESSM) regards the design, the development, and the deployment of embedded AI techniques, based on deep learning, for smart industries. Such techniques must satisfy multiple constraints such as the heterogeneity of the input sources, the limited computing power and the respect of real time constraints. The main areas we are interested in are related to anomaly detection, predictive maintenance, and security. In particular, we are designing a system, based on deep Recurrent Neural Networks, that is able to predict the time to failure of gearmotors based on data such as vibration, sound, temperature. Moreover, we are testing the adoption of object detection techniques (based on deep Convolutional Neural Networks) for improving the security of industrial plants. The activity of the node are carried out thanks to our research group and to the start-up SmartMe.io, a spinoff of the University of Messina, that is collaborating with other national and international industrial partners to test the research results in real environments. In particular, we are testing some of our techniques in the Ferrari plant in Maranello (Italy). In this sense, we believe that machine learning could lead the industrial sector into a new level by creating a further automation layer, where machines are able to understand their current state by collecting and correlating a lot of data coming from heterogeneous sensors and make decision, thus improving the reliability and the performance of the overall system.

---

Additional material

• Presentation slides: [pdf]

For more details on this presentation please click the button below: