Projects
We do not arrive with a closed solution. We listen to how you produce, what matters for quality and which data you already have — images or machine signals. Cases are split this way: Visual and Signal.
Visual Intelligence
THE EXPERT TEACHES WHAT QUALITY MEANSA few real examples, then generative AI to expand the data realistically. The expert validates. The system recognises better, even when line images are scarce.
We recognise parts picked from a cut sheet and match them to the production order, for put-to-light applications.
We recognise defects on leather fabrics, even very small ones, without replacing the judgement of those who know the material.
We recognise defects from a few sample images, without being misled by the veins, thanks to the generative-AI engine.
We inspect metal surfaces and finishes while accounting for acquisition conditions and the characteristics of the material.
Signal Intelligence
THE MACHINE SAYS WHAT IS CHANGINGThe same method, different data: vibration, current, temperature, pressure, sound, motion and PLC. Machine technicians teach what is normal. The system flags when behaviour changes.
We start from how the line runs. Then in the machine signals we look for patterns and behaviour changes that anticipate a stop, a drift or a part out of spec.
Integration
The solution can interface with PLCs, cobots, robots, AGVs and other industrial systems, including older protocols.
Acquisition, lighting, software and connection to existing machines are assessed together. The architecture follows the real constraints of production. Data stay under control: on-premise or environments defined on the project.
Ongoing research
THIS IS NOT A COMMERCIAL SUPPLYA University of Verona project with Qualyco, supported by Fondazione Cariverona. It combines hyperspectral imaging, thermography, 3D vision and polarised light with machine-learning models, for a smart inspection prototype that can be replicated across sectors.
If you produce something else, we still start from the process and the quality criterion. Feasibility is checked on the part and the machine, not on the slide.
The next step
Tell us how you produce, what you want to understand, and where quality or time is lost. Then we assess together whether you need images, machine signals, or both.