Scenaro: The pursuit of sustainable manufacturing is a key tenet of Industry 5.0 applications, focusing on enhancing process quality to enable Zero Defect Manufacturing (ZDM) and Zero Waste Manufacturing (ZWM). These goals need innovative Big Data and Machine Learning (ML) solutions to achieve production with zero defects and zero waste.

Partnership: The project leverages the expertise of a motivated partnership consisting of three advanced laboratories and three industry leaders, respectively IN4 (University of Ferrara), CIRI ICT (University of Bologna), CRIS (University of Modena and Reggio Emilia); Bonfiglioli and Carpigiani, Emilia-Romagna based companies who are global leaders in the manufacturing sectors and will provide use cases; Imola Informatica, who will provide support in the commercialization of the platform; Clust-ER Innovazione nei Servizi, who will take care of the dissemination of the project.

Objectives: The project aims to introduce a highly customizable IT platform that will extend existing IT platforms by introducing new methodologies and tools to address real-time inference, data quality from production equipment, and edge learning. It also seeks to implement MLOps solutions to keep ML models updated and efficient as production processes evolve, enabling dynamic and contextual learning.

Technologies Employed: The project will employ Big Data services and ML techniques, focusing on real-time data inference, high-dimensional data learning, dynamic MLOps solutions, and distributed learning orchestration between edge devices and the cloud.

Expected Results and Impact: By project end, a service platform will be developed to support ZDM and ZWM solutions, adaptable to various use cases within the Emilia-Romagna manufacturing sector. Expected results include new data-driven ML methodologies, a prototype DISSEM platform for dynamic Big Data analytics, an orchestration service prototype for distributed learning, and a demonstrative lab for real-time decision support services. The project aims to significantly enhance methodological and practical know-how in ZDM and ZWM solutions, fostering an increase in skilled junior researchers and advancing the state of sustainable manufacturing.

Duration: 30 months

Scenario/Context: Businesses involved in the ESCALATION project face daily optimization challenges that are crucial for accurate and effective business planning and management. Current solutions are based on monolithic algorithms tailored to single use cases and run on physical machines, leading to scalability and code reuse issues.

Partnership: The project brings together a consortium of laboratories with proven research expertise, respectively CRIS (University of Modena and Reggio Emilia), IN4 (University of Ferrara), CIRI-ICT (University of Bologna) and RE:Lab S.r.l., as well as industry leaders dedicated to advancing optimization solutions, respectively CoopService, who provides a use case connected to the transport sector and Maps Group, who provide use case in the health care sector.

Objectives: ESCALATION aims to create a micro-services integration platform for solving complex optimization problems, facilitating software reuse, accelerating evolutionary processes, and improving scalability. The project also focuses on a security-by-design approach to ensure high data protection standards.

Technologies Employed: The project leverages microservices architecture, cloud computing, fog computing, and security-by-design methodologies. It will integrate the entire computing continuum, from traditional cloud computing to edge-based solutions typical of fog computing.

Expected Results and Impact: ESCALATION's innovative framework will allow for scalable and efficient problem-solving, reducing development costs and enhancing market competitiveness. The versatile framework can be easily adapted to various sectors, making it extremely beneficial for the involved enterprises. The project aims to advance the state-of-the-art by integrating cloud and edge computing solutions, providing a flexible and open system for diverse application scenarios.

Duration: 30 months

Scenario/Context: RESIST0 aims to enhance the management of production processes by integrating shop floor Operational Technologies (OT) with data management Information Technologies (IT), particularly in Edge Computing. This integration accelerates feedback cycles within business processes in a secure environment, leveraging opportunities for servitization and optimizing production chain programming.

Partnership: The project involves a consortium including three research laboratories, CIRI-ICT (University of Bologna), IN4 (University of Ferrara), AIRI (University of Modena and Reggio Emilia) and three companies, T3LAB, SACMI, and Imola Informatica, combining expertise in digital twins, discrete event simulators, big data mining, and edge/cloud-based software systems.

Objectives: The main goal is to develop a Digital Twin-based management system for production lines and supply chains supported by cloud/edge computing, big data, and AI technologies. The project also aims to implement advanced security measures and ESG-based rating models to ensure resilient and sustainable production processes.

Technologies: RESIST0 utilizes digital twins, AI, big data, NLP for text analysis, and edge/cloud computing to create a scalable and secure framework. The project also employs advanced simulators and emulators for production line design and workforce training.

Expected Results and Impact:

• Development of a microservices-based software platform for production optimization.

• Implementation of a reference architecture for microservices interoperability and deployment in diverse cloud environments.

• Definition of security requirements and hardening/testing techniques.

• Creation of digital twins for production line simulation and training.

• Development of ESG-based supplier rating models.

• Improved production efficiency and reduced environmental impact through sustainable practices.

The project will significantly enhance the resilience and sustainability of production processes, providing businesses with advanced tools for digital transformation. The expected outcomes include reduced time-to-market, increased profitability, and the promotion of a circular economy by reusing production waste and optimizing resource utilization.

Duration: 24 months

Website: prfesr-resisto.it

Scenario/Context: The project aims to develop a Big Data analytics platform to address the specific challenges companies face in analyzing data from modern manufacturing processes. The project will tackle these issues both methodologically and technologically by developing machine learning (ML) methodologies and tools to effectively extract information from imbalanced and incorrectly labeled datasets, perform real-time inference, generate accurate synthetic datasets to address confidentiality issues and enable data augmentation, and create adaptive, data-driven manufacturing processes. The platform will be based on a microservices architecture and the MLOps discipline, utilizing mature and quality open-source technologies.

Partnership: The partnership involves several key participants: IN4  research lab (University of Ferrara); four industrial partner companies, respectively Bonfiglioli Riduttori, MEP, EMAG SU, and Marposs; and two SMEs providing technologies and/or services, i.e. Clearbox AI and DataRiver. These partners bring together extensive expertise in industrial data analysis, ML, Big Data, and manufacturing technologies.

Objectives: The primary objective is to create a customizable Big Data analytics platform to improve manufacturing process management. This includes developing ML methodologies for imbalanced and high-dimensional datasets, real-time inference tools, synthetic data generation methods, and adaptive data-driven manufacturing processes.

Technologies Employed: The project will employ state-of-the-art open-source technologies such as Apache Spark, Apache Flink, TensorFlow, PyTorch, Kubernetes, Jenkins, MLFlow, and others. These tools will support scalable, real-time, and batch data analysis, as well as the implementation of the MLOps paradigm to maintain model performance over time.

Expected Results and Impact: The project is expected to deliver significant advancements in ML and Big Data solutions for manufacturing, enhancing competitiveness and innovation in the sector. Specifically, it aims to develop new data-driven methodologies for learning ML models from high-dimensional, imbalanced, and incorrectly labeled datasets, tools for real-time/early inference in manufacturing processes, methods for generating synthetic datasets to replace real ones for confidentiality and data augmentation purposes, and the NGA4M platform for dynamic Big Data analytics services implementing MLOps. IT services developed on the NGA4M platform will be tailored to industrial use cases. Additionally, the project aims to increase methodological and practical know-how, fostering the training of junior researchers who can later be employed in the industry. The platform will be available as a customizable "semi-finished" software product, saving time and investment for companies compared to developing bespoke vertical solutions.

Click here for project website

As part of Spoke 1, Future HPC & Big Data, under Flagship 3, our research group plays an active role within the CN-HPC PNRR project, focusing on innovative solutions for Big Data and HPC applications. Within FL3, our research activities are pivotal. A key research area is the creation of federations of Kubernetes computing clusters to accelerate and scale HPC processing. High impact research conducted in this area involves developing optimization frameworks that can intelligently select the most appropriate clusters for HPC workloads, based on factors such as resource needs (CPUs, GPUs), availability, and cost.

More specifically, the BDCC Lab collaborates with Politecnico di Torino and the University of Bologna to develop Optimuum—an advanced solution for optimizing the orchestration of cloud-native applications across federated Kubernetes deployments in multi-cloud environments.

Websites:

https://pnrr.inaf.it/progetto-hpc/ 

https://www.supercomputing-icsc.it/spoke-1-future-hpc-big-data/