VISIO-STOCK: picking gun with embedded AI
It verifies by camera that the picked product is the right one at the moment of the pick. Voice guidance and optimized routes, with all the AI on the device: no cloud, no subscriptions.
The problem: the error the scanner can’t see
In manual picking, the most expensive error is the mis-pick: grabbing the wrong product with a valid barcode.
The mis-pick
The scanner gives the green light, but the product isn’t the one the order calls for. The error goes unnoticed until the customer receives it: returns, reshipping and logistics costs.
Current solutions don’t cover it
PDAs and voice picking verify the code, not the product. If the label is valid but sits on the wrong item, the system never notices.
The solution: verify the product, not just the code
A gun with embedded AI that accompanies the operator throughout the entire pick.
Smart guidance
Voice and arrows guide the operator through the warehouse along an optimized S-Shape route.
Visual verification at the pick
The moment the product is picked, the camera checks via visual embeddings that it matches the declared SKU.
Everything on the device
100% embedded AI: no cloud, no subscriptions, and no data leaving the warehouse.
Traceability and auditing
Every discrepancy is logged and audited per SKU.
Technology
Industrial hardware and proven software, ready for industrialization.
- Raspberry Pi 5 / CM5
- Hailo NPU
- Basler industrial camera
- Visual embeddings (Edge AI)
- Offline-first WMS-lite
- ERP sync via MQTT
- Odoo integration
Project status
Software
Offline-first WMS-lite with ERP sync via MQTT, guided picking, visual verification and auditing. Over 100 automated tests.
Working demo
Being built with RobotUNO on a Raspberry Pi 5, Hailo NPU and Basler industrial camera.
Industrialization
CM5 carrier board design v1.0 approved, pending industrialization, CE/RED certifications and ruggedization.
VISIO-STOCK is looking for an investor or industrial partner
The goal: industrialize the hardware —CM5 carrier, CE/RED certifications, ruggedization— and fund a pilot with end customers through Odoo partners.
RobotUNO only facilitates contact between the project owner and interested investors: it does not raise or manage funds, nor does it provide investment advice. Investing in early-stage projects involves risk, including loss of capital. More information.
Who’s behind it
Ferran Palacín
Urkolítics — industrial edge computer visionIndustrial engineer with over 20 years of experience at Air Products and a Master’s in Data Science from UPC. With Urkolítics he develops industrial computer-vision solutions on the edge.
Have a similar project?
We develop and manufacture electronic products end to end: from prototype to mass production.