The Future of Industrial Automation: Unlocking the Power of Virtualization at the Edge with Arm-Based Systems

In the modern industrial and energy management landscape, embracing cloud-native principles is more than just a forward-thinking strategy—it’s a necessity. As industries push toward more efficient, scalable, and sustainable solutions, industrial automation vendors has been at the forefront, exploring how to shift from traditional, hardware-centric systems to a more flexible, software-defined approach. Central to this transition is the use of virtualization to enable real-time, mixed-criticality operations at the embedded edge.

Partnering together, 施耐德电气 , 凯捷咨询 , Arm , and Witekio developed the Software-Defined Industrial System (SDIS), a proof of concept that highlights the transformative potential of these technologies. The journey was filled with valuable learning experiences, and the initial results are extremely promising. At the heart of this effort lies a bold vision: to prove that cloud-native techniques can drive industrial automation to new heights.

Building on a Strong Foundation: The Arm/Linux Ecosystem

Arm SystemReady IR certification ensures that Schneider’s software can easily be ported to different hardware platforms. This flexibility reduces the complexity of hardware-software integration, enabling faster time-to-market and a more scalable development process. Testing across two platforms—Xilinx ZCU102 and Raspberry Pi4—demonstrated the power of using certified hardware to create versatile, portable solutions.

A Vision for Software-Defined Industrial Systems (SDIS)

The SDIS is designed to push the boundaries of what’s possible in industrial automation, particularly at the embedded edge. Using the XEN bare-metal hypervisor to run multiple virtual machines, the SDIS can effectively manage the real-time and best-effort workloads that industrial applications demand.

Two specific workloads were designed to test the system’s capacity: a real-time workload for intelligent electronic control and a best-effort workload using machine vision, powered by 凯捷咨询 . The real-time workload ensures that critical tasks are executed with low latency, while the best-effort workload, leveraging machine learning (ML) algorithms, handles more compute-intensive tasks, such as processing video streams from multiple cameras.

Despite the ongoing fine-tuning, the SDIS has already shown impressive results in terms of flexibility, dynamism, and sustainability. These characteristics underscore its potential to transform industrial operations across a wide range of sectors.

Dynamic, Flexible, and Sustainable Results

The SDIS proof of concept has been particularly notable in three key areas:

1. Dynamic Operation: The system supports real-time and best-effort workloads that can operate simultaneously in a dynamic relationship, by consuming the underlying hardware capacity in different ratios. For example, multiple cameras can be switched on or off as needed, without disrupting the overall performance targets. The SDIS meets the demands of industrial automation even when real-time and non-real-time tasks change periodically.
2. Flexibility: Arm SystemReady IR-certified hardware reduces complexity when mixing hardware and software components. This means that the baseline functionality of mixed criticality can be ported to other applications in sectors like healthcare, telecommunications, and robotics. By focusing on software rather than hardware, industrial automation providers can create more modular, flexible solutions that are faster to deploy and easier to scale.
3. Sustainability: Operating at the embedded edge, the SDIS reduces reliance on cloud resources, saving energy and lowering operating costs. By processing data locally, the system avoids the energy-intensive task of transmitting data to remote servers. Moreover, the SDIS can be enhanced with the right orchestration tools to balance and scale the workload mix when needed, enforcing specific policies to achieve an overall better power profile of the system.

The Benefits of Virtualization in Industrial Automation

The SDIS project has already demonstrated several real-world benefits that could revolutionize industrial automation:

• Reduced latency: Virtualization often introduces latency, but the SDIS incorporates optimizations like context switching and pass-through networking to minimize delays, making the system more responsive.
• Enhanced security: By running workloads in isolation through virtualization, the SDIS enhances data security. Workload-specific data remains separated, reducing the risk of unauthorized access or data modification.
• Decreased reliance on cloud resources: The SDIS is capable of running compute-intensive applications, such as AI/ML, directly at the edge. This reduces dependency on the cloud, saving operational costs and enabling greater innovation.
• Real-time decision-making: The PREEMPT-RT option in Linux enhances the system’s ability to respond quickly to real-time events, crucial for mission-critical industrial applications.
• Faster time-to-market: SystemReady IR-certified hardware simplifies development and deployment, allowing Schneider Electric to bring new solutions to market faster than traditional approaches.

Simplified deployment and maintenance: The use of virtualized workloads makes it easier to scale and maintain industrial systems over time. New workloads can be introduced, and existing ones updated, without the need for hardware redesigns.

Capgemini played a pivotal role in the development and testing of the Software-Defined Industrial System (SDIS), working closely with Schneider Electric, Arm, and Witekio to bring cloud-native principles to industrial automation. Their involvement in this collaboration focused on integrating smart edge and IoT services, as well as machine learning (ML) and artificial intelligence (AI) algorithms to enhance the system’s capabilities at the embedded edge.

In this project, Capgemini's expertise in ML/AI algorithms was critical to optimizing real-time workloads alongside compute-intensive tasks, such as machine vision. The challenge of running these workloads concurrently, without degrading system performance, was a core focus of the SDIS experiment. Capgemini’s team worked on refining the AI-driven video monitoring system (machine vision), leveraging Arm’s Neural Network (ARMNN) inference engine to efficiently run multiple cameras in the test scenario. This ensured that the system could handle real-time performance demands in resource-constrained environments.

Our deep experience with ML/AI algorithms and smart edge solutions helped demonstrate that virtualization could indeed support mixed-criticality operations on embedded edge systems. By enabling workloads to be isolated and managed efficiently, our teams helped prove that this approach could deliver the necessary real-time responsiveness and reliability for industrial automation. Their work also showcased how AI and machine vision could be applied in a scalable, energy-efficient manner, empowering industries to harness advanced technologies at the edge.

The Proof of Concept also helps in our work building edge orchestration systems – enhancing them with the knowledge of traits like workload criticality, power budgets, industrial edge hardware’s AI capabilities, amongst others.

Capgemini's role in the SDIS experiment underlined its ability to deliver innovative solutions that enable real-time, mixed-criticality operations, bringing cloud-native principles to the embedded edge, which is expected to reshape how industries approach digital transformation.

Capgemini's expertise extends beyond industrial automation, encompassing a broader vision of "software-defined everything" (SDx) and cloud-native architectures across multiple domains, such as telecommunications, media, technology (TMT), energy, utilities, and industrial sectors. As businesses increasingly shift toward digital solutions, Capgemini is at the forefront of enabling cloud-native technologies that underpin real-time, mission-critical applications, revolutionizing how industries function. In industrial automation, the collaborative development of SDIS represents a leap forward. SDIS is proof that cloud-native principles can be applied to embedded systems at the edge, allowing businesses to take advantage of dynamic, flexible, and sustainable operations. Capgemini’s extensive experience in implementing cloud-native and SDx architectures plays a crucial role here, enabling clients to adopt modular, reusable software components for real-time, mixed-criticality systems.

The Path Forward: Unlocking New Business Models

The SDIS represents a new approach to industrial automation that has the potential to unlock new business models for industrial automation players like Schneider Electric. By adopting a software-first mindset, they can create modular, reusable software components that are portable across different hardware platforms. This not only makes the development process faster and more efficient but also allows Schneider Electric to offer more scalable and flexible solutions to their customers.

Cloud-native techniques are already proven in IT, but Schneider Electric is showing that they can bring similar benefits to the industrial edge. Over time, this software-defined approach could enable industrial automation players to develop commodity products that can be applied to a wide range of sectors—from factories and supply chains to hospitals and telecommunications.

A Bold New Future for the Industrial Edge

The SDIS proof of concept marks an important milestone in the evolution of industrial automation. By leveraging cloud-native techniques, Arm-based architectures, and a software-first development approach, Schneider Electric is demonstrating that industrial systems can be more dynamic, flexible, and sustainable.

This is just the beginning. As Schneider Electric continues to refine the SDIS and explore its full potential, it’s clear that the future of industrial automation lies at the intersection of virtualization, the embedded edge, and cloud-native development.