Steel and metal
Coils, sections and billets: heavy goods, with no label, stacked on the floor. The case where the visual identifier shows its worth most.
Ruby Neural
An empty slot in the racking that nobody has seen.
The real state of the warehouse from what is already being recorded: what is there, where, and since when.
The system says one thing. The racking says another.
The management system knows what there should be in each position. What there really is, you find out by going down to look, and that is why stocktaking happens twice a year.
Between one stocktake and the next, the gap grows: a badly placed pallet, a return that was not registered, a slot that has been empty three weeks and that the system still counts as full.
And the moment it gets found out is the worst one: when somebody goes to fetch it.
The aisle camera sees the whole racking run.
You mark the positions that have to be watched on the image. From then on you know which are occupied, which are empty and since when.
That does not replace the system: it cross-checks it. When the system says full and the camera sees empty, the discrepancy comes out with the photo.
And the figure is from every day, not from twice a year.
What it tells apart.
- Occupied and empty positions in racking, and since when.
- Its own visual identifier for items with NO label and no code.
- Stack height and stacks over the limit.
- The volume of bulk material in a pile, a hopper and a yard.
- The state of stacked material: fallen, leaning, badly strapped.
- Goods in the aisle, out of their location.
- Manual picking checks: whether what was picked is what was ordered.
Lo que Ruby Neural entiende
4 detecciones grabadas.
Logística, carga y almacén.
Ruby
Ruby
Ruby
Ruby
Cada una es una grabación del sistema funcionando. Pasa el ratón por encima —o tabula hasta ella— para verla en movimiento.
A real case.
A spare parts warehouse had a recurring stock discrepancy in one particular aisle. It was put down to picking errors.
The visual record showed something else: the pallets in that aisle got moved to the next aisle every time a large delivery came in, and nobody registered it. It was not a picking error: it was a lack of space in the inbound marshalling area.
Where it is used.
Paper and cardboard
Reels in the yard, stack height control and the state of the wrapping.
Construction
Stockpiles of aggregate, cement and precast units, with volume measured between weighings.
Agri-food
Silos, bulk bags and seasonal pallets, with levels and rotation by batch.
Pallet warehouse
Occupied and empty positions in conventional racking.
Picking
Empty pick locations and replenishment outstanding.
Inbound and outbound marshalling areas
Pallets piling up, how long they have been there and how full the area is.
Bulk materials
Pile level, hopper filling and floor area occupied.
And a warning about the numbers.
Anybody can promise that this cuts stock discrepancies by a percentage. We will not, because it depends on the state the warehouse is in today and on what gets done with the alerts.
What can be said is what gets measured: how many discrepancies appear a day, how long they take to resolve and how many reach the stocktake unresolved. Those three figures, measured before and after, tell you whether it works.
And if after three months they have not come down, the problem was not visibility and you have to look somewhere else. That has to be sayable too.
Where you start.
- One aisle or one area, not the whole warehouse.
- Two weeks measuring without alerting anybody, to tune the thresholds.
- Compare with the manual count of that area: that is where you see whether the system gets it right.
- Only then, connect the alerts and extend to the rest.
Stacked goods: what does not fit on racking.
Not all of a warehouse is numbered slots. There is material stacked on the floor: sacks, coils, seasonal pallets, sections, tyres. There is no location to look up there, there is a pile and the memory of whoever built it.
From the camera the pile gets measured: how many levels it has, whether it is within the permitted maximum, whether it is leaning and whether the material on top is the same as the material underneath.
Stack height control is not tidiness for its own sake: it is the cause of a share of warehouse accidents, and it is one of the things checked by eye and out of habit.
Bulk materials and yards.
Measuring how much material there is in a pile is surprisingly hard. It gets estimated by apparent volume, weighed on the way out, or surveyed topographically twice a year.
With the camera you get a daily estimate of the pile's volume, and above all how it moves: how much it has come down this week. For a yard of aggregate, scrap or biomass, that curve is the stock figure that does not exist today.
It does not replace the weighbridge. It gives you a number between one weighing and the next, which is where there is none today.
Manual picking.
In manual picking, the error is not found in the warehouse: it is found at the customer. And by the time the return arrives, reconstructing what happened is impossible.
By visually checking what is picked against what was ordered, the error shows at the time and gets corrected before it goes out of the door.
And when a complaint still comes in, there is an image of what went into that box.
How it goes into the system you already use.
None of this sets out to replace the warehouse management system. What it does is send it what it sees: this location is empty, this stack is over height, this pick does not add up.
It arrives as a discrepancy with its photo and its time, in the same place where the team already works. If it ends up on another screen that has to be looked at separately, nobody looks at it.
Casos grabados
Funcionando, en sitios de verdad.
Gestión inteligente de estanterías en almacén con Infinity Neural
Ruby Neural: Revolucionando la seguridad y eficiencia en muelles de carga
Ruby Neural: Revolucionando el almacenaje inteligente de materiales a granel .
What it does not do.
The visual identifier is not a barcode: it does not say which product it is, it says that THAT item is the same one that came through the door two hours ago. To know the product code you still need the barcode or the delivery note.
And on high racking, one camera covers two or three levels reliably. The ones above need another angle, and that means more cameras.
And what sits alongside it.
Other things the same family understands, on the same cameras.
What people ask us.
▸Do we have to replace the cameras we already have?
Almost never all of them, and sometimes none. What decides it is not the brand or the megapixels: it is the angle, the distance and the light. A modest camera well placed works better than a good one looking from the wrong spot.
The first thing we ask for is an image from the camera exactly as it is. With that you can see whether it works, whether it has to be moved or whether another one is needed. And if another one is needed, we say so then and not after signing.
▸Where is the video processed?
Wherever it needs to be. It can be on a machine on your own site, with the image never leaving it, or on a server. On sites where the video cannot leave for reasons of policy or regulation, it is processed inside, full stop.
What goes out in that case is not image: it is the alerts and the figures.
▸And data protection?
None of this identifies people. It tells apart figures, postures, objects and vehicles, not faces or names.
Even so, a system that processes images of a workplace has its obligations: informing people, defining what it is used for and how long things are kept. That gets sorted at the start, with whoever handles data protection at the company, and it is not paperwork left to the end.
▸How long does it take to get it working?
It depends on how many cameras and on whether they are already fitted. With existing cameras that are well aimed, a pilot area — an aisle, a marshalling area — is configured in days and left measuring without alerting anybody for a week or two.
That period is not red tape: it is what tells you whether the marked positions are right and how many false alerts there are. A system switched on for real on day one is usually switched off on day three.
Rolling it out to the rest of the warehouse comes afterwards, with the criteria already tuned.
▸And if the warehouse layout changes?
You mark the positions on the image again. Nothing has to be reinstalled or reprogrammed: it is drawing again, and your own team can do it.
In seasonal warehouses, where the layout changes every season, that is the difference between the system being used and it being abandoned at the first change.
▸How many cameras are needed?
Fewer than people assume for the basics and more for the fine detail. Counting how full an aisle is and spotting empty slots on the first two levels is done with one camera per aisle; telling one position from another up the height takes two or three.
We look at it with the layout and a photo. And if the answer is that twenty cameras are needed for what you want, we say so beforehand.
▸Is it any use for doing the stocktake?
It helps a lot and it does not replace it. The official count still needs its procedure.
What changes is that, when the stocktake comes round, the discrepancies are already located and dated instead of all appearing at once.
▸Does it replace the warehouse management system?
No, it cross-checks it. The system knows what there should be; this sees what there is. The difference between the two is the figure that does not exist today until the stocktake.
It is sent as a discrepancy to the system itself, with the photo, so that somebody resolves it the same day.
▸And goods with no label?
That is the case it solves best. There are goods that cannot be labelled: bales, coils, sections, irregular items, packaged bulk material. All of that is tracked today with a piece of paper and the memory of whoever moved it.
Ruby generates its own identifier from what it can see: size, shape, colour and the combination of features that makes it different from the ones next to it. From then on that item can be followed round the warehouse without anything having been stuck on it.
It is not a barcode and it does not say which product it is. It says that the item in position C-14 is the same one the seven o'clock lorry unloaded.
▸How many positions does one camera cover?
It depends on the height of the aisle and on the distance. On conventional racking, one camera per aisle usually covers two or three levels reliably; the ones above need another angle.
We look at it with a photo of the aisle before saying a number.
▸Does it work with the warehouse running?
Yes, and that is how it gets installed. Nothing has to be emptied and the operation does not stop; the setup is done on the live image.
Get started
Do you want this on your site?
This is already recorded working. What is left is to see whether your camera allows it: where it is fitted, what angle it has and what can be seen from there. If it will not work, we will tell you.