VRTLST Edge v2 ships with on-device inference at 38 ms

Read more

VRTLST Inspect

Catch the defect before it leaves the line

Scratches, porosity, misalignment, missing components, short fills, bad welds. Inspect makes the call inside the takt time and tells the line what to do about it — divert, flag or pass — instead of producing a report someone reads tomorrow.

INSPECTING SKU 44-921 · WELD SEAM 1,284 / hr
FAIL 2 defects · confidence 0.96 · diverted to rework

What it does

Manufacturing defect detection

01

Decisions inside the cycle

A verdict in under 50 ms, wired to the PLC over digital I/O or OPC UA. Fast enough to divert the part, not just record it.

02

Operators teach it

A one-tap correction on the HMI when the model gets it wrong becomes training data for the next round. No data-science ticket required.

03

Traceable by serial

Every unit keeps its image, its verdict and the reason for that verdict, stored against the serial number for as long as you need it.

How it works

From a camera to a decision

Sub-millimetre surface inspection at line speed, with a decision the PLC can act on.

  • Pass / fail in under 50 ms, in-cycle
  • Learns new defect classes from operator feedback
  • Writes straight to MES, SCADA or the PLC
Book a demo
Mount or reuse a camera
Existing line cameras usually work. Edge devices cover new cells and harsh environments.
Teach the defect
Label examples of the failure mode. Inspect learns the class and the tolerance band around it.
Wire the verdict to the line
Digital I/O and OPC UA for the PLC, REST and MQTT for MES, SCADA and historians.
Review and retrain
Escapes and false rejects surface in Command, feeding the next model version.
<50 ms in-cycle pass / fail decision
1,284 units inspected per hour
99.2% defect recall in production
OPC UA verdicts straight into the PLC

Specifications

The technical detail

What your integrator will ask for on the first call.

Decision latency
<50 ms in-cycle
Throughput
up to 1,800 units / hour / line
Smallest defect
sub-millimetre, optics permitting
Line interface
digital I/O · OPC UA · REST · MQTT
Traceability
image + verdict + reason, per serial
Retraining
from operator HMI corrections

Getting started

How a Inspect deployment goes

The same shape every time, whichever product you start with.

  1. Week 0

    One hour of footage

    Send us recorded video from a single line. We come back with what the model found in it — before anyone signs anything.

  2. Week 1–2

    One line, one model

    We stand up a pilot on the line that hurts most, using the cameras already mounted. Operators start correcting it on day one.

  3. Week 3–4

    Wired into the line

    The verdict moves from a dashboard to the PLC, the MES or the alerting channel your shift leads actually watch.

  4. Month 2

    Second line, second site

    The model and its thresholds travel. Command stages the rollout so a bad version never reaches more than one line.

  5. Ongoing

    It gets better

    Corrections accumulate, drift gets flagged, and new defect classes are a fine-tune rather than a project.

Questions

What people ask about Inspect

Something not covered? Talk to an engineer.

Can it keep up with our line speed?

A verdict lands in under 50 ms, which sits inside the takt time on every line we have deployed on. If your cycle is faster than that we will measure it during the pilot before committing.

What happens on a false reject?

The operator overrules it on the HMI. That correction is logged against the unit and queued as training data, so the same false reject gets less likely rather than becoming folklore.

Does it replace our existing vision system?

It can, but it does not have to. Inspect often runs alongside a fixed-function system, taking the defect classes that were too variable to write rules for.

How do we validate it for the quality system?

Every unit keeps its image, verdict and reason. Command exports the full record, and model versions are pinned and auditable per line.

Keep exploring

The rest of the stack

See the whole platform
Guard

Every zone watched. Every rule enforced.

Draw a boundary once. Guard enforces it on every shift, on every camera, without a human staring at a wall of screens.

Explore Guard
Watch

Know what's happening the moment it happens

Continuous monitoring across every line and every site, with alerts that reach the right person in seconds.

Explore Watch
Edge

The camera that thinks for itself

A sealed edge-AI device that runs the full VRTLST stack on-board. Single PoE cable in. Decisions out.

Explore Edge

Get started

Point it at a camera. See what it sees.

Send us one hour of footage from one line. We'll come back with what the model found in it.