Hall 7, Booth 380
Advantech
Industrial AI edge and software-defined automation
What you will see
Our vision AI demo at the edge: InSiSoNet finding parts on a live camera, with every result ready for the PLC.
Meet InSiSo at SPS 2026
Machines that act need to see. InSiSoNet, our proprietary on-device vision architecture, finds parts on a live camera and hands every result to the PLC, with no cloud at runtime and no extra PC. See it on two booths in Hall 7.

Physical AI starts with perception
SPS 2026 puts industrial AI at the centre of the fair. Before a machine can pick, sort, assemble or inspect on its own, it has to see: on the machine, in real time, on the silicon it already has.

Physical AI
Every part found and located in machine millimetres, then handed to the PLC or robot as a tag, like any other sensor.
Edge AI
InSiSoNet runs on the device, on CPU, Hailo-8 or NVIDIA GPU. No cloud round-trips, no extra PC.
Software-defined automation
Visual Detector installs on ctrlX OS like any other app, and is licensed and updated as software.
Two booths in Hall 7
Hall 7, Booth 380
Industrial AI edge and software-defined automation
What you will see
Our vision AI demo at the edge: InSiSoNet finding parts on a live camera, with every result ready for the PLC.
Hall 7, Booth 450
ctrlX AUTOMATION, the open automation platform
What you will see
InSiSo Visual Detector, the vision app for ctrlX OS: running on ctrlX CORE, with every result on the ctrlX Data Layer.
Messe Nürnberg, Messezentrum 1, 90471 Nürnberg
Plan your visit on the SPS websiteTrain it in ModelBench, run it with Visual Detector

Core Engine
A proprietary object detection model, architected from scratch to run inside silicon
Learn more
Vision AI Development Workspace
Label a few images, train a detector on your own classes and export it for the controller.
Learn more
Runtime Vision Intelligence Engine
Runs the model on the machine's own camera feed and hands each result to the PLC.
Learn moreBuild once. Scale everywhere.
From your parts to a result the PLC reads
Pictures of your parts, on your line, under your lighting.
Label a few images in ModelBench and let it label the rest.
Train a detector on your own classes, in the browser or on one PC.
Export the model for the controller: ONNX, or compiled for Hailo-8.
Load it into Visual Detector on ctrlX CORE or ctrlX IPC.
Find and locate each part on the live camera, in machine millimetres.
Publish every result to the ctrlX Data Layer, for the PLC to read like any other sensor.
Save pictures from the line and retrain when parts or lighting change.
Bring your use case to Hall 7
One model architecture, trained on your parts.




Talk to us about your next machine
Add vision to the machine you already sell, trained on your parts and your lighting.
Inspection and assembly checks on the line, without a vision team.
One model layer, many targets: vision for your ctrlX and edge AI projects.
Intelligence Inside Silicon.
Tell us about the line and the controller. We will show it working on your parts.