Transportation & Rail

Track-caused incidents rarely come from an unknown defect. They come from a known class of defect that sat low on a list too long to walk. Vegetation encroachment is the same: a corridor survey that takes weeks per subdivision produces a violation list that is out of date before the work orders are written.

In the depot, a hot bearing or a dragging brake gives a thermal signature days before it fails, and in the field a signal aspect or a switch position can be wrong for longer than anyone would like to admit. The data exists. The problem is that it lives in cameras and arrays nobody has time to watch.

RailGuard detects vegetation encroachment and clearance violations (structure gauge, overhead) from corridor imagery, maps each with GPS, scores severity and generates the work order. TrackSentinel classifies cracks, wear and ballast fouling from geometry car or inspection vehicle imagery, prioritizes by severity and traffic volume, and exports inspection logs aligned with FRA Part 213, with GPS, timestamps and inspector override on every record.

FleetTherm reads depot-mounted thermal arrays for wheel, bearing and brake anomalies, with component-level tracking and a lead-time estimate so the intervention can be scheduled, not forced. SignalEye verifies signal aspects and switch positions from wayside cameras and raises faults through the PTC integration.

Accuracy is stated, not implied. FleetTherm runs at 92 to 96 percent on bearing and wheel thermal anomalies, every alert carries the thermal image, and the false-positive rate is tunable by the maintenance team. RailGuard is trained on leaf-on and leaf-off imagery, with seasonal baselines set in the proof of concept and drift checked quarterly.

Wayside cameras (visible and thermal) with Jetson or x86 edge compute, depot-mounted thermal arrays, and a cloud or on-prem analytics tier. Class I railroads run fully on-prem with no cloud dependency, Class II and III carriers usually take the cloud option, and hybrid edge-plus-cloud covers the rest.

These systems add to your HBD and WID detectors; they do not replace them, and we would not recommend that anyone try.

Proof of concept in 2 to 4 weeks, pilot in 6 to 8, full deployment in 60 to 90 days. First value: 4 to 6 weeks for right-of-way monitoring, 6 to 8 weeks for track defects.