Surface inspection finds a chipped edge or an off-dimension face. It cannot find a cavity, a crack or a void inside the brick — and an internally flawed brick that looks perfect is the one that fails in the wall. The line pairs full-geometry laser inspection of all six faces with an acoustic integrity test that knocks each brick and classifies the sound.
Bricks are produced in enormous volume and inspected, historically, by eye and by sample. That catches some of the external faults — a chipped corner, an off-dimension face, poor levelling — but it says nothing about the inside, and it only looks at the bricks that happen to be sampled.
There are two different failure classes, and they need two different physical principles. External faults are geometric and can be measured with light. Internal faults — cavities, cracks, voids and density differences — are invisible to a camera or a laser, and the traditional way to find them is to break the brick, which is destructive and only tests the sample.
And the part is heavy, abrasive and dusty. Handling it — picking, spacing, aligning, reorienting and sorting — is a robotics problem in its own right, and it has to happen at a rate that keeps up with production rather than sampling it.
The ones that decide whether the line can replace sampling with full inspection, stated before any of them had an answer.
A brick can be dimensionally perfect and internally cracked. Surface and laser inspection cannot see a void, so a second, different physical principle is needed to test the inside.
Levelling and dimensional accuracy have to be checked on every face, which means reorienting a heavy brick under inspection — repeatably, and without losing throughput.
The only way to keep the brick is to test it without breaking it. Acoustic resonance — knock and listen — works, but turning a sound into a reliable verdict needs a trained model, not a threshold.
Pick, space, align, tilt and sort dusty, heavy bricks repeatably, without the wear, jams and misplacement that heavy parts cause.
The laser and acoustic blocks are marked because the line’s whole point is to test both what you can see and what you cannot — geometry with light, internal integrity with sound. The sort is marked because a brick passes only if both inspections pass.
The scope of the build, from feasibility to an installed line, rather than a capability list.
Rarely the subsystem that sounds difficult. Written out because a reader facing the same line gets more from this than from a list of what went well.
The acoustic signature separates a sound brick from a cracked one, but only after a model is trained on enough good and bad samples. A fixed threshold does not generalise across brick types, sizes or moisture, so the decision had to be learned.
Tilting a heavy brick to expose each face to the laser, repeatably and quickly, is a robotics problem that sets the throughput of the whole line.
The brick has to survive the test. That ruled out anything that marks or stresses it and put acoustics first, with ultrasound, X-ray or thermal held as fallbacks for the defect types acoustics does not resolve.
Dust and weight wear grippers and jam conveyors. Picking, aligning and sorting the parts reliably over a shift took more attention than the inspection physics did.
A brick passes only if both geometry and integrity pass. Combining two independent inspections into a single, unambiguous sort — pass, rework or scrap — is its own piece of work.
External geometry and internal integrity, where manual inspection could only judge the outside — and only on the bricks it sampled.
Every brick tested and kept, rather than a sample broken to infer the rest.
Full-geometry laser inspection with a tilting robot, not a single-face spot check.
Good and non-good separated by robot into their own carriages, with the defect data kept so trends across a run are visible.
Customer projects are presented at property, capability, outcome and integration level. Customer names, internal architecture, confidential deliverables and commercial terms are not disclosed. Where a figure would identify a customer, it is omitted rather than approximated. More detail is available under a non-disclosure agreement, within the limits the customer has agreed.
Every item links to its own page, with characteristics, applications and maturity stated honestly for that item.
The geometry inspection stage.
SOLUTIONRobotic handling and sorting.
PLATFORMThe machine vision platform.
AICombining optical and acoustic evidence.
CASE STUDYOptical inspection and sorting.
KNOWLEDGEHow the handshake works.
Because a camera or laser measures the surface and the shape, not the inside. A brick can be dimensionally perfect and still hold a void or a crack, and the only way to find that without breaking the brick is a physical test of its internal integrity — here, an acoustic resonance test read by a trained model.
Because levelling, dimension and surface defects can appear on any face, so a single-face spot check misses them. A robot tilts each brick to expose every face to the laser in turn, rather than assuming the unseen faces match the visible one.
The brick is struck and the sound is captured, and a model trained on known good and defective bricks classifies the signature. A fixed threshold does not generalise across brick types or sizes, which is why the decision is learned rather than set by hand.
It is removed by a robot into a separate carriage for rework or disposal, and the defect is recorded so trends across a production run — not just individual rejects — are visible.
Written for engineers. Share it with one.
Tell us the specification, the constraint and the deadline. Programmes that cross silicon, radio, embedded and AI are where Faststream is strongest.