Embedded AI voor machines en apparaten - FHI, federatie van technologiebranches

Seminar Embedded AI for machines and devices

Embedded AI transforms machines and devices into smart, self-learning systems that optimize processes and make decisions autonomously. During the Embedded AI seminar, you'll discover in three lectures what's already possible today and what opportunities the future holds. You'll gain insight into current AI developments and see concrete practical examples such as image processing for quality control and real-time analysis of sensor data. You'll also learn which technical choices are essential for successful implementation, from hardware to software architecture. This session offers machine and equipment manufacturers practical tools to effectively integrate AI into their systems.

Faraday Hall 9 | Thursday, September 24, 1:00 PM – 2:30 PM

From sensor to control: AI in the process industry beyond the model

Most AI initiatives in the process industry fail not because the technology is lacking, but because they start with the algorithm rather than with the batch that is at risk of going wrong. AI is not a ready-made product that you place alongside your process; it is a tool that only delivers value when it is anchored in your process data, your domain knowledge, and your way of working.

In this session, I will show what that means in practice. How to turn raw sensor data—with all its gaps, noise, and varying measurement frequencies—into something usable. Why the step from “we have data” to “we act on it” is often underestimated, and which choices regarding data streams, storage, and modeling make the difference. And why the distinction between applying existing tools and building something that truly fits your installation is crucial to whether it sticks.

No hype and no generic promises, but concrete examples from our work in the industry, including what didn't work. You will see how an iterative approach, with room to learn, leads to solutions that operators trust and that actually have an impact on the process.

Speaker: Erwin Haas, Landscape AI

Remove AI from the cloud!

The rapid growth of advanced AI language models is promising, but also potentially very dangerous and unsustainable.
Our carefully generated sustainable energy is being burned to an ever-increasing extent in enormous data centers.
A partial solution to this problem is the local application of AI. Within a defined context, specific AI algorithms deliver much better performance with less computing power.
In this presentation, the latest developments in this field are shown. Modern microcontrollers are already being equipped with a dedicated “Neural Processing Core” to classify sensor data locally, for example. This prevents vast amounts of data about us and our behavior from ending up in data centers, with all the associated potential dangers.
So remove AI from the cloud!

Speaker: KITT Engineering, Andries Lohmeijer

Hardware selection for edge AI systems in practice

How do you choose the right hardware for an intelligent system at the edge of the network? In this presentation, we show how sensors, signal processing, available data, and computational requirements together determine which embedded hardware is suitable for an edge AI solution.

Using two practical examples—a medical measurement system and a vision system—we discuss how design choices at the system level impact model complexity, energy consumption, latency, and implementation on the final platform.

This provides accessible insight into the technical considerations behind the step from algorithm to working embedded / edge system.

Speaker: Demcon, Bas Vet