Revolution at the Edge: PLC-FPGA Synergy for an Ultrafast and Sustainable Industrial AI

Modern industry faces a critical challenge: the need to make intelligent decisions in real time directly on the production floor. Artificial Intelligence solutions based on the cloud often encounter barriers of latency and connectivity. To overcome this, at DeustoTech we are leading a project that redefines the limits of Industrial Edge-AI.

The “Dúo Dinámico” of Advanced Automation

We are carrying out a pioneering integration that combines the best of two technological worlds. On the one hand, the deterministic reliability and proven robustness of the Siemens S7-1215C automaton, a standard in industrial control. On the other hand, the immense flexibility and parallel processing capacity of the FPGAs of the AMD Kria KV260 family.

We call this combination the “dynamic duo” of IA Industrial. By combining these technologies, we can run complex Deep Learning models directly on the machine. The result? A drastic reduction in latency, bringing response times down to microseconds, something fundamental for the control of critical processes, advanced robotics and industrial artificial vision.

The Research Challenge: Performance vs. Consumption

However, taking artificial intelligence to the extreme (Edge) is not just about computing power. As part of the work of our research group at DeustoTech, we are addressing the core of the problem in embedded systems: the delicate balance between computational performance and energy consumption.

A massive industrial deployment of AI will not be viable if it is not sustainable. Therefore, our research focuses on the development and application of advanced techniques to optimize Deep Neural Networks (DNNs). Through strategies of compression, quantization and efficient design of algorithms, we ensure that the inference is not only ultra-fast, but also:

Energy efficient: Minimizing the consumption of watts for each inference made in the FPGA.

Thermally stable: Reducing heat dissipation, crucial in closed industrial environments.

Highly scalable: allowing its integration in diverse production lines without requiring oversized electrical infrastructures.

Looking towards Industry 5.0

The work we are developing from DeustoTech not only solves a current technical problem, but also lays the foundations for Industry 5.0, where machine intelligence collaborates efficiently, quickly and, above all, sustainably.