But first, the basics: for Syassen, two approaches are particularly relevant: classical machine learning, which is used in products such as AI-supported cameras, and generative AI (GenAI), which is becoming increasingly important. Sick has already developed products that use deep learning models to recognize patterns, for example in quality control or parcel sorting. These solutions make it possible to operate industrial applications more efficiently and with greater accuracy.

One specific example is camera systems used in logistics centers. Here, integrated AI models ensure reliable barcode recognition, package measurement and sorting optimization – a crucial factor in times of increasing shipping volumes.

Challenges for image processing

Although machine vision is considered a commodity by many in the industry, Sick still sees potential here. The challenge is to develop robust solutions that work reliably under demanding industrial conditions. According to Syassen, the focus here is less on developing models and more on combining proven hardware with application-specific AI software. The so-called “industry-grade AI” is the focus here.

AI and safety: a complex interplay

A key topic of the discussion was the integration of AI into safety solutions. Classic safety systems are based on deterministic approaches, while AI is built on probabilistic models. These different approaches require innovative solutions to meet the high demands on functional safety.

One example is the Safe Brake Assist system developed by Sick and used in road construction machines. With the help of laser scanners and machine learning, the system can reliably detect people and prevent accidents. The technology was developed to work reliably even in challenging conditions such as steam or fog. The system is certified and offers a new dimension of safety in industrial environments.

And GenAI? Syassen mentioned application examples such as support in the search for lost luggage or the optimization of internal processes. Sick has identified over 60 GenAI applications, some of which are already in use, for example in software development or in the analysis of invoices. Efficiency through specialized AI models

For Sick, the future lies in smaller, specialized models that require less computing power and energy. According to Syassen, the focus here is not so much on developing models as on combining proven hardware with application-specific AI software. The so-called “industry-grade AI” is the main focus here.