Manipulating deformable objects (DOs) that change shape during operations poses a challenge for robots due to the complexity of modelling and predicting the object’s state. This challenge is critical for the food industry, where collaborative robots will assist workers in processing and packaging food products. Most meat products are three-dimensional DOs (3D-DOs) composed of biological tissues, which give them viscoelastic and anisotropic properties that make their behaviour highly unpredictable and challenging to model. To address this challenge and minimise food cross-contamination, dual-arm robots and non-actuated end-effectors, such as chopsticks, are used to provide more flexibility, especially when working with objects in confined spaces, and reduce tool actuation time. Our goal is to develop a methodology for manipulating predominant convex 3D-DOs composed of biological tissues on a planar surface, altering their position and orientation while minimising their deformations. To accomplish this, we propose bimanual motion primitives utilising a novel geometric approach. Effectiveness is measured by the Intersection-over-Union ratio and object centroid error metrics through 180 chicken-breast, 36 pork-loin-steak experiments, and 96 comparative experiments with heuristic methods for rigid objects.

This paper highlights the development and implementation of a digital twin for a pilot within the AGILEHAND project, aimed at improving planning and resource management in food processing. By integrating real-time data, the digital twin enhances forecast accuracy, optimizes raw material usage, and ensures efficient scheduling. The system models the physical production line, including multiple processing stations and conveyors, using an open source discrete-event simulation. Additionally, key elements such as products and operators are modelled to reflect real-world interactions, allowing for detailed analysis of production line behaviour across various scenarios. Moreover, Artificial intelligence plays a critical role in optimizing operations, managing order processing, and addressing inventory overlap to ensure smooth transitions between production phases. This comprehensive approach facilitates precise production tracking and adaptive scheduling, ultimately improving operational efficiency and resource utilization.

Automated fish grading and packaging, critical in the seafood industry, have not been simultaneously addressed by prior works using machine vision and robotics. This letter presents a novel proof-of-concept robotic vision system for automatic, size-based fish grading and packaging. Our system classifies frozen fish steaks into two size grades and localizes them on a conveyor belt for robotic pick-and-place via a specialized end-effector. Experiments achieved a grading accuracy of 87.6% and a robotic packaging rate of 87%, demonstrating the potential of vision-guided robotics for automated food quality inspection and handling.

This research investigates the application of computer vision for rapid, accurate, and non-invasive food quality assessment, focusing on the novel challenge of real-time raspberry grading into five distinct classes within an industrial environment as the fruits move along a conveyor belt. To address this, a dedicated dataset of raspberries, namely Rasp Grade, was acquired and meticulously annotated. Instance segmentation experiments revealed that accurate fruit-level masks can be obtained; however, the classification of certain raspberry grades presents challenges due to color similarities and occlusion, while others are more readily distinguishable based on color. The acquired and annotated RaspGrade dataset is accessible on Hugging Face at: https://huggingface.co/datasets/FBK-TeV/RaspGrade.

This paper presents the development of a dynamic control system designed to sort citrus fruits based on external defects identified through multispectral imaging. The proposed system combines PID controllers integrated through the OPC-UA protocol to facilitate real-time adjustments and optimize batch classification. By dynamically controlling classification thresholds, the system maximizes output within customer-specified tolerances, enhancing batch utilization and quality. Implementing Node-RED as an OPC-UA client enables seamless integration and real-time data processing, ensuring efficient and responsive control. Additionally, Docker containerization is employed to streamline deployment and scalability, enabling flexible and resource-efficient implementation across various operational environments. The system also introduces virtual classification tiers that dynamically adjust sorting thresholds, allowing for better utilization of each batch while maintaining adherence to strict quality specifications. Experimental evaluations demonstrate that the system significantly improves sorting quantities, optimizes classification efficiency, and reduces waste by adapting to varying conditions in real-time. These enhancements ensure high standards of quality and efficiency in citrus fruit classification while supporting sustainable agricultural and industrial practices through automated inspection.

In food production, numerous factors significantly influence both the growth and quality of products. Predicting product quality and quantity in an early stage and precise way is particularly difficult. For accurate order planning, reducing waste, and ensuring customer satisfaction, production planners need to know the provided product quality and quantity from the suppliers in a precise way. To address this need and support planners in their daily decision-making processes, an approach has been created that involves the development and application of machine learning models aimed at initially predicting product quantity and subsequently product quality in two distinct phases. The developed machine learning pipeline focuses specifically on the case study of raspberry production. By utilizing production and field data, alongside publicly available weather data and synthetic datasets, various machine learning models were tested and validated.

The globalization of agroindustry in the twenty-first century has introduced new challenges related to ensuring product safety in international organizations. Mandatory traceability regulations are inherently scarcely effective in such context, since their validity is limited to the national boundaries, while new and more effective solutions can be provided by voluntary traceability systems. Since traceability systems ultimately rely upon the establishment of appropriate product related information flows among the supply chain operators, their integration with interoperable industrial automation systems may offer substantial advantages. The spread of cyber-physical systems within the industry 4.0 paradigm thus constitutes an ideal technological layer for such development, although the lack of an integrated framework involving product traceability and process control still hampers the practical development of such systems. This paper focuses on discussing the complexities related to the integration of traceability requirements in interoperable automation systems, from the perspective of the existing standardization frameworks in agroindustry, thus providing managerial insights for industrial applications.

The food manufacturing sector increasingly requires advanced technologies for production management and process optimization. In particular, the development of Digital Twins (DT) enhances synchronization between logistics and production by integrating innovative technologies such as Big Data, IoT, and Artificial Intelligence. These technologies enable process optimization, reducing wait times and improving resource utilization. In the food sector, this translates in minimizing food waste and ensuring product freshness. This study focuses on optimizing production management and human resource allocation in collaboration with an Italian company specializing in small fruit processing as part of the AGILEHAND European project. The optimization approach includes improving forecasting techniques through AI-based algorithms, optimizing workforce allocation and streamlining production workflows to improve adaptability to fluctuating demand. The DT serves as a decision-making tool by analysing both real-time and historical data, simulating various operational scenarios, and determining optimal production configurations using a non-real-time simulation module. The results demonstrate the system accuracy in meeting demand deadlines while optimizing workforce requirements, leading to more efficient resource management. By leveraging DT technology, the system enables proactive adjustments in production schedules and workforce distribution.

The Production Logistics system is generally a large-scale complex system with various operational phases and management levels that must integrate. In the specific context of soft and deformable food products, the core of AGILEHAND European project, this complexity increases further due to challenges related to the handling and movement of such items. Efficient coordination of production and logistics phases becomes crucial to ensure product quality, prevent losses, and optimize the entire process. In this article, focus will be placed on a data-driven framework for the automated generation of simulation models, serving as the foundation for digital twins in intelligent factories within the previously mentioned sector. The proposed framework represents a multi-layered data-driven system designed for real-time/near-real-time simulation, planning and synchronization of production and logistics systems during line reconfiguration. The digital model forms the basis for a digital twin with simulation and optimization capabilities, designed to facilitate decision-making at various management levels in the production and logistics process and control activities such as changes, maintenance, quality and safety. Exploiting information provided by the Enterprise Traceability system, the digital twin aims to establish a real-time/near-real-time information flow. This flow enables accurate capturing of dynamics occurring in the physical layer and effective assessment of their negative effects on the overall operational state of the system in the digital layer. In this context, the use of the digital twin is intended to simplify and expedite the reconfiguration of production and logistics systems. This is achieved through the early detection of system design or process sequence through cross-sectional simulation.

The automation of grading, handling and packaging of soft and deformable food products in the industry necessitates precision in manipulation, adaptive gripping, and real-time monitoring to ensure uniform quality and prevent damage. These aspects are the objectives of the European project AGILEHAND. This project aims at developing advanced technologies as a strategic instrument to improve flexibility, agility and reconfigurability of production and logistic systems of the manufacturing companies. In this context, this paper discusses the development of a data-driven framework for automated generation of simulation models for the realization of the digital twins in food factories. The core of this framework involves discrete event simulation: by real system data it is possible to streamline the reconfiguration of production and logistic systems. During this process, it is also possible to promptly identify design or process sequence problems at an early stage through cross-domain simulation. Firstly, this framework supports the project manager in scheduling production by optimising resources and minimising the order of Estimated Time of Arrival. Then, during the system monitoring phase, it simulates new scenario to find the proper corrective action when a significant deviation from the plan occurs. In the end, the accurate capture of dynamics arising in the physical layer allows an effective evaluation of their negative effects on the system’s overall operational state. This approach supports informed decision-making in planning and controlling logistics, production, and associated activities, providing a comprehensive framework for enhanced operational efficiency.

Vision and Robotic technologies are progressively becoming ubiquitous for automating and digitizing quality control in the food industry. This paper examines the crucial role of advanced automation technologies, including versatile or dedicated robotic systems, specialized end-effectors, machine vision, and efficient material handling systems, which collectively enhance food processing efficiency. This paper aspires to provide a high-level technical review on the crucial role of advanced automation technologies, including versatile or dedicated robotic systems, specialized end-effectors, machine vision, and efficient material handling systems, which collectively enhance food processing efficiency. While the manuscript aims to document the various automation sub-systems utilized generally in food processing, it places a particular emphasis on the primary processing phase of food production. Most food products in the primary processing phase exhibit a plethora of complex physical properties and manipulation conditions, making it difficult to reliably automate the various processes. This research aims to outline the contemporary advances and requirements for integrating various automation technologies, to enhance the efficiency and precision of primary food processing. Furthermore, it aspires to serve as a valuable, up-to-date survey and analysis of the latest advances in automation and vision technologies and their capability to automate a food processing line.

Market shifts and changing consumer demands highlight the challenges of traditional mass production techniques. This workshop proposes an Artificial Intelligence-integrated system with a multi-layer Digital Twin for optimized food production, adapting to product characteristics and facilitating real-time monitoring. Traceability services maintain product and process information, complemented by Digital Twin services projecting potential scenarios. Data-driven AI models optimize decision-making, from production layout adjustments to operational enhancements. Throughout it, human oversight is ensured using interactive dashboards, integrating technology with expertise. Implementation involves monitoring variables, managing model complexity, conducting analyses, applying knowledge effectively, interacting with stakeholders, and ensuring interoperability across functionalities.

We propose 6DGS to estimate the camera pose of a target RGB image given a 3D Gaussian Splatting (3DGS) model representing the scene. 6DGS avoids the iterative process typical of analysis-bysynthesis methods (e.g. iNeRF) that also require an initialization of the camera pose in order to converge. Instead, our method estimates a 6DoF pose by inverting the 3DGS rendering process. Starting from the object surface, we define a radiant Ellicell that uniformly generates rays departing from each ellipsoid that parameterize the 3DGS model. Each Ellicell ray is associated with the rendering parameters of each ellipsoid, which in turn is used to obtain the best bindings between the target image pixels and the cast rays. These pixel-ray bindings are then ranked to select the best scoring bundle of rays, which their intersection provides the camera center and, in turn, the camera rotation. The proposed solution obviates the necessity of an “a priori” pose for initialization, and it solves 6DoF pose estimation in closed form, without the need for iterations. Moreover, compared to the existing Novel View Synthesis (NVS) baselines for pose
estimation, 6DGS can improve the overall average rotational accuracy by 12% and translation accuracy by 22% on real scenes, despite not requiring any initialization pose. At the same time, our method operates near real-time, reaching 15f ps on consumer hardware.