Seminar Artificial Intelligence - FHI, federatie van technologiebranches

Branche

Wednesday, September 24, 10:00 AM – 11:30 AM (Faraday seminar room)

AI transforms embedded systems by making them smarter and more autonomous. During the seminar on Wednesday 24 September, you will gain insight into what is already possible and what the future holds in three 25-minute lectures.

We start with an overview of current and upcoming AI capabilities. We then show concrete examples of AI applications in embedded systems with a focus on practical benefits and challenges.

This includes: image processing for quality control, real time analysis of (sensor) data and automatically taking actions based on those insights. Finally, we zoom in on the technical choices you need to make to implement AI effectively: hardware requirements, software architecture and optimizations for reliable, efficient systems.

An inspiring session for anyone who wants to use AI in embedded technology – today and tomorrow.

10:00 – 10:25

Make an impact with AI

Many AI initiatives fail because they start with technology, not with the real problem. But AI is not a ready-made product; it is an extensive toolbox with which you can achieve a lot, provided you make the right choices. In this session I will therefore provide a clear overview of what AI means in concrete terms (and especially what it does not mean), explain why it is important to distinguish between applying existing solutions such as ChatGPT and developing AI solutions specifically for your organization, and share practical tips for an effective start. No hype, but concrete examples from our practice, where you can see how an iterative approach, with room to learn, will lead to real impact.

Erwin Haas, Landscape AI

10:30 – 10:55

Technology and infrastructure for embedded AI

In this presentation, Mark Boer of Aemics will guide us through the technological foundation and infrastructure required to successfully integrate AI into embedded systems. He will discuss various hardware approaches, such as the use of specialized AI chips versus integrating algorithms on standard microcontrollers. The software side will also be addressed: how do you optimize an AI model so that it can run reliably, efficiently, and in real time? Mark will discuss the considerations involved in choosing an architecture, including performance, energy consumption, cost, and maintenance. A practical case study will be used in a vision application, providing a common thread. This presentation offers insight into both the technical and strategic choices involved in implementing embedded AI on the edge.

Mark Boer, Aemics

11:00 – 11:25

AI case study – customer case where the benefits of AI are clearly visible and AI is already on the device.

A robot that inspects the water supply network, immediately signaling in an ICU why a premature baby stops breathing, and simulating CT scans of all conceivable forms of malignant lung tumors to better diagnose cancer. These are just a few examples of projects we are working on to apply AI in our daily lives. AI is not a distant prospect and is not limited to CoPilot, ChatGPT and a greater chance of being hacked. In this presentation I will show how we tackle these challenges and why these projects are possible now, while this was not the case 10 years ago.

Marnix Zoutenbier, Demcon

Read the interview with Marnix and his colleague about smart electronics and AI.

Artificial Intelligence Seminar

AI in embedded systems: current possibilities, future opportunities