Object detection is a critical task in various real-time applications, including surveillance, autonomous vehicles, and industrial automation, especially within the emerging paradigms of Industry 5.0 and Industrial Internet of Things (IIoT). However, deploying such algorithms on resource-constrained embedded devices poses significant challenges due to their limited computational power and memory resources. This paper presents a comparative evaluation of YOLOv5 and YOLOv8, two state-of-the-art object detection algorithms, on different embedded hardware platforms. Leveraging insights from existing research, we aim to address this work by conducting comprehensive experiments to assess the performance of the proposed algorithm in terms of detection precision and inference speed on each hardware platform. The analysis provided encouraging results and revealed different behaviors, showing the importance of GPUs also for embedded systems.

The introduction of the Industrial Internet of Things (IIoT) is dramatically changing the concept of manufacturing, ensuring better production flexibility, efficiency, safety and security. In this scenario, Functional Safety Networks are ever more deployed, being networks that allow to implement functional safety systems, integrated and cooperating with factory communication infrastructures that are ever more characterized by the deployment of wireless communication systems. Unfortunately, nowadays, the lack of safety protocols targeted for wireless networks represents a bottleneck in the novel smart factory development process. Thus, functional safety over wireless is becoming a hot research topic. In this paper, we address the adoption of Wi-Fi to implement functional safety networks by exploiting the black channel approach, which is at the basis of the most popular functional safety protocols designed for wired networks. In practice, with such an approach, the safety protocol is not aware of the underlying communication system. We focus on a specific protocol, namely FailSafe over EtherCAT (FSoE) and investigate its behavior over Wi-Fi. To this aim, we developed an experimental set-up and conducted several tests to adequately assess safety, reliability and timing performance. Specifically, we addressed the achievable Safety Integrated Level (SIL), the number of network re-initializations and the message delivery times. The analysis provided encouraging results and revealed different behaviors concerned with the use of different transport layer protocols (TCP and UDP) that suggest interesting future activities.

In recent years, Factory Automation is evolving towards the so-called Industry 4.0, and the creation of a smart factory ecosystem comprising of ubiquitously interconnected objects, namely the Industrial Internet of Things (IIoT), is gaining much research interest. This paradigm aims at developing new smart technological equipment and protocols, thus providing interconnection among "factory objects" anywhere and at any time. In this context, people and machines have to safely cooperate and a high level of protection needs to be guaranteed for both operators and the surrounding environment. For this reason, safety systems, aiming at decreasing risks and failure probabilities, are nowadays of uttermost importance. Several Functional Safety communication protocols have been developed during these years pointing to increase data integrity and guarantee protection in a safety system. Popular examples are Fail Safe over EtherCAT (FSoE), ProfiSAFE, and OPC-UA Safety. These protocols, al-though conceived for wired networks, can be in principle adopted also by wireless communication, as they are developed by using a black channel approach. Nevertheless, the implementation of these protocols over different wireless networks is challenging as they might not ensure the required Safety Integrated Level (SIL). This paper, moving from the aforementioned observations and the need for wireless solutions in the IIoT context, focuses on proposing a possible implementation of FSOE over Wi-Fi, running UDP at the transport layer. In particular, by using suitable experimental outcomes, an OMNeT++ simulator has been calibrated, thus enabling the possibility to analyze the proposed protocol in wide industrial systems.

Modern industrial manufacturing plants, especially those using coordinated electrical drives with strict timing requirements, make extensive use of real-time communication networks. These systems, typically, are based on various topologies, include diverse protocols, and connect devices from different manufacturers, which may make them difficult to study, plan and optimize. As a solution, the adoption of digital twins allows to simulate such systems under various operating conditions in a low-cost and zero-risk environment. In this paper we address the digital twin of a networked electrical drive system, focusing on the real-time communication network used to connect the drives. In particular, we describe the simulation model of Profinet IO RT Class 1, implemented as an extension of the INET library of OMNeT++. Moreover, we present the outcomes of the tests carried out on a prototype simulated network and compare them with those of the equivalent real one.

Wireless technologies play a key role in the Industrial Internet of Things (IIoT) scenario, for the development of increasingly flexible and interconnected factory systems. Wi-Fi remains particularly attracting due to its pervasiveness and high achievable data rates. Furthermore, its Rate Adaptation (RA) capabilities make it suitable to the harsh industrial environments, provided that specifically designed RA algorithms are deployed. To this aim, this paper proposes to exploit Reinforcement Learning (RL) techniques to design an industry-specific RA algorithm. The RL is spreading in many fields since it allows to design intelligent systems by means of a stochastic discrete-time system based approach. In this work we propose to enhance the Robust Rate Adaptation Algorithm (RRAA) by means of a RL approach. The preliminary assessment of the designed RA algorithm is carried out through meaningful OMNeT++ simulations, that allow to recognize the beneficial impact of the introduction of RL with respect to several industry-specific performance indicators.