manufacturing. The control performed by sensors is more efficient for in‐line monitoring procedures, where all sensors are synchronized in order to provide the best production setting of the whole supply chain. All the phases of the supply chain are important to trace. The main parts to trace in the production processes are: (i) warehouse, (ii) production lines, and (iii) logistics. In all these parts, robotics in general improves the processes, by increasing production volumes and by assisting human work. This kind of “joint collaboration” decreases the production errors and consequently the waste materials and related costs. Other technologies such as AR [5, 6] are used for human resources training during production processing, by increasing the worker skills and supporting workers to follow interactively and continuously the production. Augmented reality aided manufacturing (ARAM) is another important topic supporting production quality [5] by means of the programming of machines, robots and production tools, by managing logistics, and by checking assembled products in the whole supply chain. AR is adopted also in manufacturing as a dynamic authoring tool monitoring simultaneously the production activities of several workstations [6], for telerobotics controlling robots from a distance, for waste reduction in production activities, for assembly support, for remote maintenance, and for computer‐aided design (CAD) applications [7]. In the Industry 4.0 scenario, AI can furthermore improve the industry production efficiency. AI algorithms are mainly indicated for machine predictive maintenance [8, 9] and for assisted production, where machine working operations are properly and automatically set in order to avoid failures [10], by decreasing or stopping machine in cases of alerting conditions. IoT sensors are very important for control and actuation thus enabling totally automated processes. A broad use of IoT sensing is related to image vision [11, 12] including IRT [13], and temperature and humidity sensors. Moreover, accelerometers provide supplementary information about anomalous vibrations indicating a possible system failure, and other sensors can be applied depending on the manufacturing process to be controlled. IoT signals are processed by AI algorithms to predict the machine status in self learning modality: by analyzing historical data, the AI algorithms create the training models to test for prediction. The AI improvements represent mainly the passage from Industry 4.0 to Industry 5.0 facilities adapting automatically the production with high level efficiency, and optimizing the production processes which are previously simulated. The flexibility of the production is due to the correct choice of the sensor network architecture, of protocols and the possibility to optimize the different layers of the whole communication system of the company. A correct design of the information system allows the disposal of a modular network open to vertical and horizontal integrations introducing innovative tools and algorithms addressing the automatic production control. The layers where it is possible to operate for a flexible production are the input/output (I/O) layer, the user interface layer, the gateway layer, the IoT middleware, the processing layer, and the application layer.
1.1.1 Sensors and Actuators Layer: I/O Layer
The I/O layer is the first layer related to the production field controlled by sensors. The process of machines can be changed by actuation commands provided by the processing layer. The actuation commands must ensure the production synchronization of the whole production lines managing different production steps. In this layer, IoT devices are very important for the accuracy and reliability of the performed measurements. The data sampling is essential for a correct monitoring procedure. When the sensors control different production process steps, it is fundamental to configure and to synchronize all the sensors of the same production line. The IoT technologies are defined for the specific production process to monitor. For example, if the process is fast, it is important to select an image vision technology having a high frame rate, or sensors having a sampling time “following” the production velocity. The technologies for industrial image vision converting light into electrons are charge‐coupled device (CCD), complementary metal oxide semiconductor (CMOS), indium antimonide (InSb) infrared (IR) detectors, indium‐gallium‐arsenide (InGaAs), germanium (Ge), and mercury cadmium telluride (HgCdTe) sensors. Table 1.1 shows the working wavelengths of the IR technology.
Table 1.1 Spectral ranges of infrared technology.
| Infrared technology | Spectral range (μm) | References |
|---|---|---|
| InSb | 0.6–5 | [14] |
| InGaAs | 0.9–1.7 | [14] |
| Ge | 0.8–1.6 | [15] |
| HgCdTe | 1–9.5 | [14] |
Sensor networks are designed after an accurate analysis of the production processes, thus suggesting the correct configurations and connections of possible gateways, routers, and of device combinations. Sensor networks are implemented for point to point, star, extended star, bus, or mesh configuration. In Figure 1.1 are shown the different main network configurations. The design of the network is an important step for the realization of the correct network. The spatial allocation of the production machines and the workflow of the production define the best configuration. The network layout changes with the sensor system: the star or mesh network is typically adopted for sensors, besides the bus layout is suitable for production line connections and for the information system. By considering for example a photovoltaic camp with a high number of panels, it is preferable to realize a ring type fiber optic network linking all electrical string panels. The network also assumes a hybrid configuration, especially when a new network is added and linked to an old one. Table 1.2 lists the main advantages and disadvantages of the different network layouts.
Figure 1.1 Example of network configurations: (a) point to point connection; (b) bus line; (c) ring layout; (d) star connection; (e) tree layout; and (f) node meshing configuration.
Table 1.2 Advantages and disadvantages of network typologies.
| Network type | Advantages | Disadvantages |
|---|---|---|
| Star | The star network manages the whole network by a single node behaving as a master node. Each node of the network can be added, removed, and reconfigured by ensuring the network operations. Network simplicity. Easy identification of errors | For a failure of the central node the whole network is out of order. Bandwidth limitation |
| Bus | Low cost and simple layout. Connection with a simple coaxial or RJ45 cable | For a failure of the bus the whole network is out of order. Additional nodes decrease network velocity. Single direction transmission mode (half duplex) |
| Ring | Bidirectional transmission mode for dual ring typology (full duplex) | Half duplex modality for basic ring configuration. Transmission security (if a node fails the network stops operating) |
| Tree |
|
