Wrong shelf, lost item
A package in the wrong location is effectively gone until someone stumbles over it.
Warehouse vision AI · Built in Lörrach
Inventory that sees itself.
BARCONA AI detects, reads and assigns every barcode in your warehouse, and flags what doesn't add up. It runs on a camera mounted to the forklifts you already have, or on an autonomous robot.
Live · Detection Unit
BARCONA AI core
Detect · Read · Assign · Check
Output: results.xlsx
The problem
Warehouses lose track of what they already own.
Stock gets put in the wrong place, labels get damaged and manual counts lag days behind reality. Each mistake costs time to search, re-pick and correct.
Counting takes people
Hand scanners and stocktakes tie up staff and only give you a snapshot, never the current state.
Labels fail
Damaged, covered or missing barcodes break systems that rely on scanning alone.
BARCONA AI
One image in. A structured result out.
Our software core turns camera images and video into barcode values, location assignments and clear status flags, ready for Excel or your systems.
Simplified illustration of a BARCONA AI result. BARCONA AI also recognises the arrow on the shelf label, so it knows which level the label belongs to. Labels and IDs are fictional.
Two ways to deploy
Same AI core. Two ways to the shelf.
Use the forklifts you already run, or send an autonomous robot on defined routes. Or combine both.
Detection Unit
Test setup in preparation
Detection Unit
A camera and compute module on existing forklifts. It captures inventory during normal operations, needs less robotics and is built from standard components.
Explore the Detection Unit →TraceUnit-1-30xIllustration
Prototype
TraceUnit-1-30x
An autonomous robot that runs its own scan missions on defined routes, independent of forklift traffic, with barcode, AI vision and LiDAR.
Explore the TraceUnit →How it works
Three sensors. One picture of your floor.
Each sensor covers for the weaknesses of the others. Together they give a reliable, continuously updated picture of every package. The animation shows the sensors our robot would incorporate, step by step.
Sensor: Barcode
Illustration
Illustration, not a product design. It shows the sensors our robot would incorporate and what the system can do with them as a whole. In practice they sit together on one device; which sensors are used and how they are mounted can be adapted and discussed for your site.
01 · Scan
Reads every label it passes
A camera reads the barcode on every label as the unit drives along the aisle. Nobody has to point a hand scanner.
- Continuous scanning while moving
- Identifies item and package ID
02 · Recognise
Sees what labels can't tell it
A camera with AI visual recognition checks what a package actually looks like, so damaged or hidden labels don't leave gaps.
- Visual cross-check of scanned items
- Works when labels are unreadable
03 · Map
Knows exactly where everything is
LiDAR maps the space around the unit, giving every detected package a precise position on a live floor map.
- 360° detection range
- Real-time location mapping
04 · Flag
Spots mistakes immediately
When something is in the wrong place or labelled incorrectly, the system flags it, so your team fixes it now and not at the next stocktake.
- Misplaced package detection
- Mislabel detection
Simulation
Watch a scan mission.
We test navigation and scanning in simulation before anything runs on real hardware. Here's a simplified example mission: the robot stops at each rack and angles its camera upwards to read every shelf level, right to the top.
Simulation · example hall
Simplified demo
- Barcodes scanned
- 0/16
- Mission time
- 0.0 s
- State
- Driving
Scan log
LiDAR: 360° distance scan
Camera: field of view
Not scanned
Reading
Scanned
Built for
Wherever stock moves fast.
Storage & fulfilment
Continuous visibility over racks and floor space, without scheduled stocktakes.
Distribution hubs
High throughput means high error risk. Catch misplaced items before they ship wrong.
Research environments
A flexible platform for testing autonomous inventory and sensor-fusion ideas.
Progress
Building in public.
We're a young team and we share where we are, honestly. Follow along as BARCONA moves from first idea to the warehouse floor.
Read the updates →The first idea
Stocktakes are slow, manual and out of date the day they finish. We started asking how a camera could do the counting instead.
A concrete concept
The idea became a plan: the TraceUnit, an autonomous robot with camera, barcode reading and LiDAR that scans racks on its own.
The first software
The core architecture for sensor input, data processing and mapping, plus a first camera setup that picks out barcodes in live images.
Datasets & camera control
We started building our own datasets of real warehouse labels, and the camera learned to pan, tilt and zoom towards what it detects.
Camera testing
From early April we tested the camera chain against our datasets end to end, then added a tilt scan that reads a rack level by level.
Warehouse simulation
A simulated hall lets us test navigation and scan routines before they run on hardware.
Our own detection model
We trained a recognition model on our own warehouse images.
BARCONA AI software core
The software core is done: it turns warehouse images into barcode values, shelf locations and status flags.
Detection Unit test setup
A first forklift-mounted test setup, followed by tests in real warehouse conditions.
The team
The people behind BARCONA.
KB
Co-founder · Lead Software Engineer
Kenan Busch
Leads the software architecture of BARCONA AI and its integration into our hardware.
PW
Co-founder · Lead AI Engineer
Philipp Wirth
Leads the AI behind BARCONA: the vision models that recognise and read barcodes in real warehouses.
Work with us
Help us put BARCONA on a real floor.
We're looking for warehouses and logistics teams who want early access, direct influence on the product and a close working relationship with the people building it.