INDUSTRY 4.0 IN THE AGRI-FOOD SECTOR: INNOVATIVE SENSORS AND SMART LOGISTICS TO SUPPORT THE SUSTAINABILITY OF THE SUPPLY CHAIN

The agri-food sector is undergoing significant transformations in measurement technologies, with traditional laboratory-based analytical techniques increasingly being replaced by rapid, non- destructive, and environmentally friendly methods. Visible and Near-Infrared (vis/NIR) spectroscopy and imaging techniques are highly regarded for their non-destructive sampling, real- time results, and ability to perform quality controls throughout the production process. However, these techniques typically involve expensive, complex instruments often impossible to use outside of a specialized laboratory. To overcome these problems, low-cost prototypes are being created that maintain the reliability of laboratory instruments, although with significantly lower performance. The aim is to ensure high-quality products that meet customer demands for healthiness and sustainability through meticulous quality controls. Chemometrics plays a crucial role in interpreting data, predictive modeling, and variable selection, which can simplify these devices. This project aims to meet the needs of high value-added supply chains by implementing simplified spectroscopic devices utilizing a limited number of wavelengths. Over the three years, not only were prototypes of portable instruments used, but several studies also involved the use of benchtop instruments. These served as a reference for the optical acquisitions obtained using the prototypes, as well as for conducting feasibility studies and with the aim of implementing this type of instrumentation in-line within the food industry. Commercial instruments used included Aurora NIR, (Grainit), ASD Quality SpecTrek, Jaz (Ocean Optics), Corona Process® (Zeiss), and Unity® (Hellma). Data analysis followed a specific procedure involving spectra exploration, chemometrics analyses and modeling, using Microsoft Excel, Matlab® and the PLSToolbox package as the main software. A first prototype, which also led to the filing of a patent, was developed and tested on olives, tomatoes and wine grapes and it has been designed specifically for use on small-sized matrices. The optical device was used to evaluate table tomatoes' quality features and then to quantify qualitative parameters of Chardonnay grapes directly in the field. Regarding wine grapes, Life Cycle Assessment (LCA) methodology was also applied to evaluate the sustainability of the optical analysis compared to traditional laboratory analyses, demonstrating the environmental advantages of spectroscopy. Research activities continued with grape matrices during post-harvest phases, using benchtop instruments to measure key parameters at consignment. The feasibility of vis/NIR spectroscopy for rapid polyphenol content determination in grapes was assessed and the final models provided valuable information on phenolic parameters, with promising results that require further testing in wineries. Another experiment focused on evaluating the environmental and economic performance of an automated system for managing yeast nutrition and monitoring the fermentation process. In line with the principles of 4.0 winemaking, another simplified prototype was developed with the aim of analyzing non-solid matrices, such as wine, must, or other liquids, for detecting adulterations and measuring qualitative parameters. This prototype was used to monitor wine must fermentation and craft beer production, testing its applicability to different liquid matrices and fermentation types. Another experiment was conducted using benchtop instruments for the determination of acrylamide in cocoa biscuits, yielding promising results that align perfectly with Industry 4.0 and the in-line implementation of spectrophotometric analyses. A third prototype has been developed and tested but is still in the pre-prototype stage (Technology Readiness Level, TRL of 3 out of 9) and requires further experimentation before reaching the performance of the other two prototypes (TRL 4). Experiments were conducted on larger agri-food matrices, such as apples and pears, always using commercial instruments as references, given the early stage of development of the prototype. The goal was to monitor qualitative parameters and create predictive models for ripening and senescence. Visible/Near-Infrared spectroscopy is a valuable technology for rapidly, non-destructively, and sustainably assessing quality parameters of several food and agricultural products. The prototypes developed showed promise for certain applications but require further experimentation. Further development and testing are necessary to enhance the applicability of these prototypes in the agri- food sector.