Building More with Less: Applying Lego Logic to Embedded System Consolidation

Growing up, I loved building Lego. I’d spend hours making contraptions that could do it all—fly, swim, dive, and drive. The real magic of those creations wasn’t just in what they did—it was in combining everything into one. Instead of building separate machines, I had a single system that handled every job. It was more efficient, more compact, and honestly, more fun. 

That same principle shows up today in embedded systems through workload consolidation. Rather than relying on multiple pieces of hardware to handle different tasks, we can now bring everything together—running diverse workloads on one high-performance platform. It saves energy, reduces complexity, and streamlines maintenance—all without sacrificing performance. 

And unlike my early Lego inventions (which were impressive but not exactly space-efficient or durable), today’s consolidated systems are built to be lean, robust, and highly optimized. 

Less hardware, same functionality  

With workload consolidation, you don’t lose the specialized performance or expertise of individual systems. Each application remains independent, with its own resources for computing, memory, interfaces, and even graphics. They run side-by-side on one hardware platform, fully isolated, and fully functional. 

The key ingredients? 

• • • A multi-core processor (which, luckily, all modern x86 processors provide - even the most power-efficient ones). 
    • A real-time hypervisor to allocate dedicated resources to each workload and ensure fault isolation. If one crashes, the others keep going – completely uninterrupted. 
      
      

All you need is multi-core and virtualization 

That’s exactly what we demonstrated at Embedded World with our live setup: a ball-balancing demo. 

We packed multiple workloads into one setup - just like I used to do with Lego, only way more sophisticated. At the core is a COM Express Compact module based on Intel Core Ultra technology. These processors offer not just high CPU and GPU performance, but also a Neural Processing Unit (NPU) designed for local AI inference. 

Here’s what we did: 

• • • Virtual Machine 1 (AI): Used the NPU and the Intel OpenVINO toolkit to track the ball with a USB camera.
    • Virtual Machine 2 (Real-Time Control): Ran the CODESYS SoftPLC to control the x-axis of the balancing plate.
    • Virtual Machine 3 (HMI): Provided the user interface for monitoring and interaction. 

In the mood to play  

And because we were having fun, we didn’t stop there. Instead of handling the y-axis control on the same module (which could be done easily), we connected another building block: a SMARC 2.1 module based on Intel Atom technology, linked via real-time TSN running its own Ethercat instance to control the y-axis motor. With this setup the smooth real-time operation over TSN was also demonstrated. 

Even this power-optimized module had more than enough performance to run: 

• • • VM 4: A second Codesys instance for y-axis control. 
    • VM 5: A system dashboard showing real-time status and system health. 

We even considered adding a third virtual machine to host an IIoT gateway for secure cloud connectivity - but time was tight, and deadlines are deadlines. 

From Demo to Industry 

Now imagine this setup scaled to an industrial environment. With fewer physical systems to manage, you free up resources—not just for space and cost, but for creativity. You could deploy a full production cell with an autonomous, collaborative robot working in real-time alongside human coworkers. With just one module per robot, you could handle AI, real-time control, safety systems, and communication - all consolidated, all virtualized. 

And if the robot is stationary? You could run the entire cell on a single platform. 

Sound complex? It doesn’t have to be. There are application-ready hardware and software building blocks available that make system consolidation not just possible - but practical.  

We’ve got the technology. We’ve got the building blocks! Now it’s your move.  

And while you’re at it, check out some of these helpful resources we’ve compiled… 

• • • Success Story: How Ono Sokki cut system count from seven to one 
    • An introduction to workload consolidation
    • Wireless TSN: The future of industrial communication 
    • See the demo in action