Introduction: The invisible Engine of Innovation
Imagine a world where artificial intelligence, advanced research, and industrial innovation are no longer limited by computational power. This is the vision that drives modern European industry. As we move toward more immersive digital experiences, like the Extended Reality (XR) training environments developed in the MOTIVATE XR project, the complexity of the underlying technology grows exponentially.
As AI models grow more complex and data-driven experiments become the norm, access to robust, flexible, and secure computing infrastructure is no longer a luxury: it is a necessity. While we often marvel at the “front end” of innovation, the VR headsets, the smart robots, or the chat interfaces – we rarely see the engine room that makes it all possible.
This is the promise of High-Performance Computing (HPC), a notable change for industries and researchers alike. To meet this demand, BI-REX has introduced Ipazia, a new HPC infrastructure designed to bridge the gap between innovative research and real-world industrial applications. Ipazia is not just a cluster of servers; it is a shared platform that empowers enterprises and European projects to turn bold ideas into scalable solutions, faster, smarter, and together.
Extended Reality (XR) is reshaping industrial training, making complex tasks safer and more efficient. At the heart of this transformation lies MOTIVATE XR, a Horizon Europe project that combines XR, AI, and digital twins to democratise immersive training tools. But behind the scenes, these innovations require immense computational power. This is where Ipazia, BI-REX’s High-Performance Computing (HPC) cluster, comes into play.
Why a Dedicated HPC Hub Matters for Europe
In the landscape of European innovation, reliance on external or generic cloud providers can create barriers. Ipazia is conceived as a shared technological backbone for the BI-REX consortium, bringing together enterprises, research centers, and innovation partners.
Its mission is to provide accessible, high-performance computational resources that accelerate innovation across multiple domains. For a project like MOTIVATE XR, which relies on innovative simulations, having a dedicated hub is critical. The infrastructure is specifically designed to support industrial AI and data science projects, reducing AI training costs by up to 40% compared to commercial cloud solutions.
Furthermore, European-funded projects require robust and secure compute environments to ensure compliance with GDPR and Horizon Europe data sovereignty requirements. By centralising computational power and expertise, Ipazia strengthens collaboration and reduces entry barriers to advanced AI. This ensures that consortium members can focus on innovation, creating better XR tools and safer training modules, rather than worrying about infrastructure constraints.
The MOTIVATE XR Project: a European Vision
MOTIVATE XR brings together 17 partners across Europe to create an integrated XR tool suite for industrial operations. The goals include:
- Easy XR authoring tools for non-programmers.
- Affordable digital twins for real-time simulation.
- Collaborative XR environments for remote and on-site teams.
- Industrial pilots in aerospace, manufacturing, energy, and consumer appliances.
Beyond the Cloud: Performance and Sovereignty
To the general public, “the cloud” often sounds like the ultimate solution for everything. However, for specialised industrial research, it has limitations. While cloud solutions offer flexibility and on-demand scalability, Ipazia’s HPC infrastructure is specifically optimised for high-performance, research-intensive workloads.
Unlike generic cloud environments, Ipazia provides predictable performance, dedicated resources, and tighter control over data. This is a critical distinction. In a generic cloud, you are often sharing resources in a way that can lead to variable performance. Ipazia offers fixed, optimised resources for long-term use.
Ipazia: The Backbone of Innovation
The IPAZIA HPC cluster is designed with a robust and modular architecture to meet the needs of modern research and industry. Users access the system through a dedicated login node, which connects to a pool of high-performance CPUs and GPUs. This setup is managed by a workload manager, ensuring efficient scheduling and resource allocation. The cluster features both fast and large storage options, enabling seamless handling of diverse data-intensive workloads. This architecture guarantees security, stability, and high performance for all users. Furthermore, IPAZIA offers an advanced, reliable, and easily accessible platform for executing complex computational tasks. Its architecture separates access and computation functions, enhancing security and operational stability. Shared storage allows all compute nodes to access the same files, while fair use and time-linearised policies ensure equitable resource distribution. This approach enables users to estimate job costs in advance and manage their budgets effectively.
Ipazia is more than a cluster; it is a shared infrastructure designed to accelerate AI and XR projects. Unlike generic cloud solutions, Ipazia offers:
- Predictable performance for compute-intensive tasks.
- Data sovereignty aligned with GDPR and Horizon Europe standards.
- Cost efficiency, reducing AI training costs by up to 40% compared to commercial clouds.
HPC Infrastructure vs Cloud Solutions: Key Differences at a Glanc
- Performance: Where traditional cloud solutions offer variable, shared resources, Ipazia delivers predictable, dedicated resources critical for training complex models.
- Data Control: Instead of being limited by a provider’s policies, Ipazia ensures full compliance and sovereignty, a non-negotiable factor for European research.
- Latency: For local and regional access, Ipazia minimises latency, which is essential when processing large datasets for real-time applications.
- Vendor Lock-in: Ipazia utilises open, interoperable standards, avoiding the high risk of vendor lock-in often found in commercial cloud platforms.
This approach reduces latency and ensures compliance with industrial and European research requirements, making Ipazia particularly suited for long-running, computer-intensive projects.
The AI Revolution: Small Models, Big Impact
One of the most exciting capabilities of the Ipazia cluster is its ability to manage the next generation of Artificial Intelligence. One of the initial concrete applications of Ipazia has been in the training and fine-tuning of Small Language Models (LLMs).
Why “Small”? In an industrial context, bigger is not always better. These models are becoming increasingly important in industrial contexts where efficiency, specialism, and data privacy are critical. Small Language Models are compact, efficient AI systems designed for specific tasks, like analysing industrial data or automating customer support. They run faster, cost less to operate, and can be hosted securely on-premises.
Seeing the world through AI Ipazia takes this a step further with Vision-Language Models (VLMs). VLMs go beyond text; they ‘understand’ both images and text, enabling applications like automated quality control in manufacturing or interactive digital archives for cultural heritage.
In addition to text-based models, Ipazia has been employed to train VLMs capable of combining region of interest (ROI) selection on artworks and images. While this sounds technical, the application is profound. Imagine an AI that can look at a video feed of a factory floor, identify a specific machine part, understand its maintenance history, and generate a text report for a human worker. That is the power of combining Vision and Language, and it requires the heavy computing power that Ipazia provides.
Industrial and European Projects: Shared Value Through Compute Power
IPAZIA supports a wide range of industrial and research applications, including:
- Digital Twins: Creating virtual models of production plants for simulation and optimisation.
- Predictive Maintenance: Leveraging big data from IoT sensors to anticipate equipment failures.
- Computational Fluid Dynamics (CFD): Simulating fluid and aerodynamic flows, as in the OASIS project for precision medicine.
- AI training and Finetuning: Training large language models (LLMs) and neural networks on proprietary datasets, such as the Nier-Gemeb project for regulatory compliance in the railway sector.
- Computer Vision: Industrial quality control and object detection for autonomous vehicles (e.g., Boston Dynamics, MIR).
- Vision-Language Models (VLMs): Integrating computer vision and natural language processing for advanced AI applications.
- Life Sciences: Accelerating drug discovery and genomics research.
How does this translate to the MOTIVATE XR project? Ipazia is not limited to experimental use cases; it is designed for industrial-grade deployment.
Enterprises within the BI-REX consortium use the cluster to develop and validate AI-driven industrial solutions, run large-scale simulations, and prototype innovative digital services. This is where the constructive collaboration with MOTIVATE XR becomes vital.
For example, within the MOTIVATE XR project, the cluster supports the development of AI-driven virtual training environments. These are not just video games; they are complex simulations that help workers master complex industrial tasks in a safe, simulated space.
The real-world benefits:
- Safety: Workers can practice dangerous procedures without physical risk.
- Efficiency: This approach reduces onboarding time.
- Cost Reduction: It minimises the risks and costs associated with traditional training methods.
- Hybrid Human-Robot Manufacturing Pilot: BI-REX supports XR-based training for collaborative robots, enabling workers to interact with robots in virtual environments before stepping onto the factory floor.
- AI-Driven Training: Ipazia powers the training of Small Language Models and Vision-Language Models for XR applications, such as real-time quality control and immersive troubleshooting.
- Safety and Efficiency: Workers can practice hazardous tasks in XR simulations, reducing onboarding time and minimising risks.
By serving multiple partners within a common environment, Ipazia fosters knowledge exchange and accelerates collective progress. It ensures that the heavy lifting of rendering simulations and processing user data is managed securely and efficiently, ensuring reproducibility and scalability.
Collaboration Stories and Lessons Learned
BI-REX’s involvement in MOTIVATE XR highlights the value of cross-sector collaboration. Challenges like integrating XR with legacy systems were overcome through joint workshops and iterative testing, fostering a culture of shared innovation.
Conclusion: The Future is Shared and Immersive
Ipazia is more than infrastructure; it is a catalyst for the next wave of European innovation. By providing a powerful, shared, and secure HPC environment, BI-REX is not only supporting its partners but also contributing to a broader vision: a future where AI, research, and industry collaborate without boundaries.
As projects like MOTIVATE XR continue to push the limits of what’s possible, blending the virtual and physical worlds to empower workers, Ipazia will be there, powering the computations that turn visionary ideas into reality.
For enterprises, researchers, and European initiatives, the message is clear: the future of innovation is shared, and it starts here.
Authors
BI-REX Competence Center
Antonio Candido is a Technical Project Manager at BI-REX Competence Center, Bologna, Italy. With a PhD in Mechanical Engineering, Antonio specializes in Industry 4.0, enabling technologies, energy optimization and technology transfer, supporting innovation through advanced digital solutions
BI-REX Competence Center
Roberto Giudiceandrea is an Automation and Robotics Application Engineer at BI-REX Competence Center. Roberto focuses on robotics integration, virtual commissioning, and enabling technologies for Industry 4.0, with expertise in digital twins and collaborative automation solutions.
BI-REX Competence Center
Marcello Simonati is an AI Engineer at BI-REX Competence Center, specialising in the design and deployment of artificial intelligence solutions for industrial applications. Marcello’s work focuses on machine learning model development, computer vision systems, and optimisation of AI workflows within high-performance computing environments. With experience in data analysis, predictive algorithms, and integration of AI into digital manufacturing processes, Marcello contributes to projects that enhance automation, quality control, and XR-based training solutions for Industry 4.0.
