Shipping errors are prevented by checking the load where it actually leaves: at the exit gate, against the shipment list in your ERP. With vision AI, fixed cameras count the crates on a moving forklift, read every crate label and match both against the SAP shipment plan, so a missing, extra or wrong crate is caught before the truck closes its doors rather than at the customer's receiving dock.
This guide looks at why crate-level shipping errors are so persistent, where manual scanning falls short, and how shipment and crate verification with TIM INSPECT turns the loading gate itself into the checkpoint. It also covers what to plan for when you install cameras at a dock.
A typical outbound shipment in automotive supply or manufacturing logistics is made up of dozens of returnable crates. Each one carries a label with a part or crate code, and the shipment is only correct when three things agree: the number of crates on the vehicle, the identity of each crate and the plan in the system. Any of the three can drift.
Once an error passes the gate, it becomes a cost: extra transport, a short delivery at a customer who may run a just-in-time line, a claim to investigate and a stock record that no longer matches reality. And there is usually no evidence of what physically passed.
Most sites verify outbound loads with handheld terminals. The operator scans each crate label before or during loading, and the WMS or ERP compares the scans with the order. In principle this works. In practice it depends entirely on the person holding the scanner.
End-of-day counts find errors after the truck is gone. The dock needs a check that does not rely on someone remembering to do it.
In a crate verification setup built on TIM INSPECT, the gate does the checking. In the automotive supplier deployment described on the solution page, four fixed cameras watch the exit gate and the forklift passes without stopping. The flow has four steps.
TIM INSPECT receives the vehicle information and the expected crate and product list from SAP. Every pass through the gate is linked to that shipment record, so the system knows what should leave before anything moves.
As the forklift drives through, wide-angle cameras count the crates on its forks from the image. Counting is independent of label reading, which matters: a crate with a missing or unreadable label is still counted.
Cameras read the code on each crate label. TIM INSPECT compares the valid reads with the crate count and the SAP list. If four crates are counted but only three labels are valid, the pass is not verified.
An unlisted, missing or unreadable crate opens an exception. AXION, the wearable worn by the operator and the supervisor, gives a voice alert, and an alarm lights up at the gate. The correction and the repeat verification stay linked to the same event, and completion is only confirmed in SAP once every expected crate is verified and every open exception is resolved.
Each pass is stored with the crate count, the codes read, a photo and a short video. Expected, verified and missing crates sit on one screen, which gives you evidence when a customer questions a delivery. On the solution page, the deployment reports 30,000 shipments, 1.5 million crates read and 100% shipment accuracy, based on the Thread In Motion project report.
Counting and reading are different jobs, which is why the deployment uses wide-angle cameras for the load and narrow-angle cameras for labels. Plan positions around where labels face as the forklift passes, how high crates are stacked and the route forklifts actually take. A layout walk with sample crates is the fastest way to find blind spots.
Doors open and close, daylight changes and forklift headlights shine into the lens. Stable, even lighting at the gate makes reading more reliable. Check label condition in advance too: worn or dirty labels will correctly show up as exceptions, but you want that to be rare.
Decide early which record defines the expected load, how exceptions are handled and who is allowed to close them. The value of the system comes from the two-way link: the plan comes from SAP and the verified exit goes back to it.
Gate cameras will capture forklift drivers. Point cameras at the load rather than at people where possible, define retention periods for photos and video, and inform employees. Under the EU General Data Protection Regulation (GDPR) and similar laws, the purpose of recording and who can access it should be documented before go-live.
The described project runs at one gate. The same approach can be planned across several connected gates, but starting with one lets you tune positions and exception handling before scaling. If your outbound flow is pallet-based rather than crate-based, four-side pallet scanning reads the labels on all four sides of a pallet in one pass.
It is the automated check that the crates leaving a gate match the shipment plan in count and identity. With TIM INSPECT, fixed cameras count and read the crates and the result is matched with the SAP shipment list.
No. In the deployment described, the forklift is not stopped and does not wait for manual scanning; verification happens while it moves. SAP completion stays blocked until any mismatch is resolved.
No. This application verifies at crate level, matching crate counts and labels with the shipment list. Counting parts inside a crate is a separate inspection scope.
An unreadable or missing label creates an exception. If the crate count and valid reads do not match, the pass is not verified, and the correction and repeat check are tracked in the same record.
If you want to see how this would fit your crates, labels and gate layout, you can request a demo and walk through the shipment and crate verification flow with our team.