Wheels, rail, under-gear, clearance, overhead line and depot components — measured in millimetres rather than judged by eye, on moving vehicles and in the shed. Built on 3D laser profiling and imaging that we specify, install and prove against your own wear limits.
Depot, wayside and on-vehicle measurement points. The same measurement engine and record structure sits behind all of them.
Every application on this page rests on the same physics. A laser line is projected across the target; a camera at a known angle sees where that line lands; the displacement of the line in the image converts to height by triangulation. Thousands of cross-sections per second, indexed to travel, become a geometrically correct surface.
The grey dashed line is the design profile; the red line is what was actually measured. Every application below is a variation on this same three-step sequence.
Three properties make triangulation the right choice trackside. The sensor carries its own light, so a measurement taken at two in the morning matches one taken in direct afternoon sun. Nothing touches the asset, so there is no wear item and no risk of damaging a wheel or a contact strip. And the output is a dimension rather than an image, which means a maintenance rule can be written against it — a photograph of a worn flange still needs a human to decide.
The constraint that governs every design decision is travel. A wheel passing at speed presents each section of its surface once, briefly. Profile rate, exposure time and encoder indexing are therefore chosen from the line speed at the outset. Getting that wrong is not a tuning problem discovered at commissioning — it is a sensor selection problem, and by then the sensor is bought.
Each of these is an application in its own right. Most operators start with one, prove it, and add the next against the same records and the same limits.
What each system measures, what comes out of it, and where the engineering difficulty actually sits. Sensor selection, accuracy figures and installation detail follow the trial on your asset — we do not quote a number we have not measured ourselves.
Full wheel geometry on every axle of a passing train, without taking it out of service.
Two heads per rail: the gauge-side view resolves the flange face, the field-side view the tread and rim. Neither can see the whole profile alone, which is why the pairing is not optional.
The wheel is not a static target. It is rotating, translating, and moving laterally within the flangeway all at once, and the profile you want is a single radial cross-section. Encoder-locked capture is what resolves this: each profile is stamped with travel rather than with time, so the reconstruction stays geometrically correct whether the train is doing 5 km/h or 60.
The second difficulty is datum. A wheel profile is meaningless without knowing where the measurement origin sits relative to the rail. That is why the two heads are calibrated as a pair against a common reference and re-verified periodically — a drift of a fraction of a millimetre in the relationship between them appears as a systematic flange thickness error across every wheel on the railway, which is exactly the kind of fault that erodes confidence in the whole installation.
The handheld depot system exists partly for this reason. It gives you an independent measurement of the same wheel against the same limits, so the wayside installation can be verified rather than trusted.
The same measurement engine and the same wear limits, taken to the wheel in the shed.
A handheld instrument is only as repeatable as its datum. The measurement geometry is fixed by a mechanical wheel locator rather than by how the operator holds it — the fixture decides where the sensor sits relative to the wheel, so two operators on two shifts produce the same number on the same wheel. Without that locator, a handheld 3D sensor is a very expensive way to reproduce the variability of the gauge it replaced.
Continuous rail condition captured from a moving vehicle, every profile carrying a chainage position.
Wear is reported as the difference between the measured section and the design profile, not as a photograph of a worn rail.
The sensors are rigidly mounted to a bogie or trolley that is itself bouncing on the track being measured. Rail wear is a small number; carbody movement is a large one. Two things separate them. Gauge and relative rail geometry are derived within the measurement frame, where common-mode motion cancels between the two heads. Absolute geometry needs an inertial datum, because no optical sensor can tell the difference between the rail dipping and the vehicle rising.
The practical consequence is that a system quoted purely on sensor specification will disappoint. What sets useful accuracy here is the rigidity of the frame, the quality of the encoder reference and the inertial datum — the sensor is rarely the limiting element.
Full-consist imaging with trained defect models, plus 3D where the fault is a shape rather than a shade.
Only exceptions travel off site. The full image set stays in the trackside cabinet, which is what keeps the link budget realistic at a remote location.
A detection system that flags a hundred items a day where two are real will be switched off within a month, and no amount of model accuracy recovers the operator's trust once that has happened. The design target is therefore not raw detection rate but the ratio the control room can actually work through in the time available between consists.
Two decisions do most of the work. Every alert carries its source image, so adjudication takes seconds rather than a site visit. And faults that are geometric — a component hanging lower than it should — are measured in 3D rather than classified from a 2D image, because a shape gives you a number and a shadow gives you an opinion. Reserving the trained models for genuinely cosmetic or presence-type faults is what keeps the alarm list short enough to be worked.
One installation covering dragging equipment, underframe clearance and loading gauge as a measured envelope.
A beam tells you a line was crossed. A profile tells you by how much, on which vehicle, at which axle — and leaves a record for the post-event review.
A conventional dragging-equipment detector is a binary device, and it cannot distinguish a hanging brake rigging from a plume of steam, a spray of ballast dust or a bird. Every false trip stops a train. The operational cost of that, accumulated across a year, is usually a larger number than the equipment.
A profiling installation measures a surface rather than interrupting a beam, so a transient obscuration appears as a single implausible profile among thousands and is discarded, while a genuinely low component appears consistently across successive profiles at a fixed position on the vehicle. The same measurement also yields height and width against the permitted envelope, which means one installation covers three duties that would otherwise be three separate systems on three separate structures.
Contact wire geometry and pantograph condition, measured as the vehicle passes — no contact and no possession.
The strip section and the wire geometry are two halves of the same failure. Measuring both from one installation is what turns a dewirement into something predicted rather than investigated.
Residual thickness tells you when a strip is due for replacement. The shape of the wear tells you why. A groove worn at one position means the wire is not staggering across the strip as designed, and the same strip will wear through early every time it is replaced until the overhead line is corrected. Even wear across the full width means the system is behaving.
This is why the two measurements belong together. The depot sees a strip wearing quickly and replaces it; the overhead line team sees stagger drifting at a particular structure. Neither, alone, connects the two. A single record carrying both, tied to chainage and vehicle identity, is what lets one team's finding reach the other's work list.
The governing constraint here is not optical. This is a high-voltage installation, so isolation, shielding and clearance to live parts decide where equipment can physically be placed, well before any measurement consideration is discussed.
Dimensional and defect inspection of parts once they are off the vehicle.
The part is stationary, indoors, under controlled light, and already in the operator's hand. Every constraint that makes a wayside installation difficult is absent. That makes the depot cell the cheapest place to establish the things that carry over to everything else — your tolerance sets, your record structure, your operator interface, and your team's confidence in a measurement produced by a machine.
Several operators start here for exactly that reason, then move outdoors once the internal argument about trusting the numbers has already been won on parts people can hold in their hands.
Every system on this page reuses the same layers. The sensing head, the fixture and the tolerance set change from one application to the next. The synchronisation, the verdict engine, the operator interface and the record structure do not.
Running across every layer: calibration and traceability, sensor health and self-test, time synchronisation and data integrity, remote diagnostics, access control and retention. That is what makes a second and third system cheaper than the first — and what lets a wheel measured at the wayside and the same wheel measured in the shed be compared without an argument about whose number is right.
A sensor that measures beautifully on a bench and fails in month three at a wayside site is not a product. These are designed for on day one, not discovered during commissioning.
The most common way a wayside installation dies is not a broken sensor. It is a gradual one: the optical window films over with brake dust and diesel residue, the data degrades slowly, the numbers drift, and by the time anyone notices, several months of records are suspect and the maintenance team has stopped believing the system. Nothing ever alarmed, because nothing ever failed.
The answer is a self-test that measures the sensor's own condition rather than only the asset — a reference feature in the field of view, checked on every pass, that alarms when the return degrades past a threshold. It is unglamorous, it appears nowhere in a sensor datasheet, and it is the single thing that most determines whether an installation is still trusted in year three.
The same measurement capability, at whichever level of responsibility suits your programme.
We bring the sensor to your depot or your track and capture real data on your asset — no slideware and no simulated results. Accuracy figures, sensor selection, installation detail and programme timescales are issued against your application, under a mutual confidentiality agreement.
Start a railway enquiry Call +91 92723 32597