Stop the bad cut before the cutter finishes it.
Corvanis EdgeIoT measures every aluminium profile while it is being machined, decides on the machine whether it is still in tolerance, and halts the spindle in under 2 ms when it is not. Every measurement lands in Odoo as a quality record.
Book a pilot See how the digital twin works
Running today as a testbed on a 3-axis CNC machining centre for architectural aluminium profiles, built with the University of the Algarve.
Fast where it has to be. Connected everywhere else.
The edge board has two brains. A real-time microcontroller guards the machine, a neural processor judges the part, and only the verdict travels on to your ERP.
Real-time core
STM32H5 running Zephyr RTOS on the Arduino VENTUNO Q. Closes its control loop in under 20 µs and trips the hardware interlock in under 2 ms.
AI core
Qualcomm Dragonwing with a 40 TOPS NPU. Runs the vision models and the uncertainty engine on the device in under 12 ms, so images never have to leave the shop floor.
Resilient transport
MQTT Sparkplug B and OPC UA with ring buffers. Delivers in under 45 ms and keeps queuing in order when the network drops.
Odoo 20
Writes the result to manufacturing, quality and maintenance records in under 250 ms, with a signed calibration certificate attached.
Latencies are design targets for the current testbed.
What changes on the shop floor
Bad parts stop at the machine
A laser scanner checks every profile at 100 frames per second against its guard-banded tolerance. The edge board stops the spindle before scrap piles up.
Maintenance before the breakdown
An accelerometer on the spindle tracks vibration against its own baseline. When wear crosses the warning line, Odoo opens a maintenance request and reserves the spare tool.
Certificates nobody has to type
Each batch gets a digital calibration certificate in the PTB DCC format, signed on the device and attached to the lot, the work order and the delivery.
Scrap measured, not estimated
Rejected profiles are reconstructed in 3D, so scrap is booked by real volume and weight and routed back for recycling.
Energy and OEE per work order
Spindle power is sampled at 1 kHz and logged against each work order, so OEE and energy cost come from the machine instead of a spreadsheet.
Keeps working offline
If the plant network drops, events wait on the edge board and arrive in order, without duplicates, once it is back.
Your ERP becomes the digital twin
There is no separate dashboard to maintain. Each signal from the machine becomes a record your teams already use in Odoo 20 Enterprise.
| From the machine | In Odoo |
|---|---|
| Profile dimension and its uncertainty | Quality checks with a guard-banded pass or fail |
| Spindle vibration | Equipment health and maintenance requests |
| Machine state and spindle power | Work centre status, OEE and energy per work order |
| Volume of rejected profiles | Scrap moves at real weight |
| Signed calibration certificate | Attached to the lot, the work order and the delivery |
Ask about your plant in plain language through Claude, connected to Odoo over MCP, and plug in other systems through Odoo's JSON-2 API.
Start with one machine
A pilot instruments a single machining centre and connects it to your Odoo, so you can judge the results on your own parts.
Map the tolerances
We review the machine, the profiles it cuts and how they are inspected today.
Fit the edge board
We mount the VENTUNO Q, the laser scanner and the spindle accelerometer, and connect them to your network.
Connect Odoo
Measurements, maintenance signals and certificates start arriving as records in your database.
Compare
We put scrap, downtime and inspection effort side by side with your baseline.
Built at the University of the Algarve
Corvanis works from the UAlg Tec Campus in Faro and runs its research and internship programme with ISE-UAlg's degree in Systems and Computer Technologies Engineering (LESTI).
- Edge infrastructureLinux, Podman and a hardened MQTT broker for the VENTUNO Q boards.
- Vision on the NPUQuantised YOLOv8 models judging aluminium profiles on the device.
- Data into OdooPostgreSQL pipelines and Odoo modules that turn telemetry into traceable records.
- FACODI academyOur community learning platform, starting with a course on digital literacy and citizenship. Visit the FACODI courses
Show us your hardest tolerance.
We will tell you whether edge metrology can hold it, and what a pilot on your machine would involve.