RailGuard

A corridor survey takes weeks per subdivision and the list it produces is stale before the first crew is dispatched. A single scan flags everything green; it cannot tell last month's regrowth from a tree that has been leaning into the clearance since spring.

Vegetation control is a federal obligation with four specific failure modes: dry brush near structures, blocked signal sightlines, obstructed crossing sight triangles and roll-by inspection clearance. Each needs documentation with a location and a photograph, and a notebook does not scale to thousands of miles.

Cameras capture 4K corridor imagery at up to 40 miles an hour on a hi-rail vehicle, continuously from a wayside mount, or at higher speeds from a locomotive in revenue service. An NVIDIA Jetson runs the models on board; nothing leaves your network. Models trained on multi-season railroad imagery identify vegetation encroachment by type and risk, structure gauge violations against AAR Plate B-K standards, overhead and catenary proximity, and crossing sight obstructions.

The first season's surveys establish the baseline; later surveys flag deviation, so growth velocity, leaf-on versus leaf-off change, herbicide effectiveness and post-storm windfall are read as differences rather than as a wall of detections. Violations push to Esri ArcGIS as a layer with severity and heatmaps, and to Maximo or SAP PM as work orders with GPS, photos, severity and a recommended action.

Vegetation within 10 feet of the centerline is detected at above 95 percent with under 5 percent false positives after seasonal calibration; in dense canopy the figure is 85 to 92 percent, and every violation carries the image so the field crew confirms before dispatch. Structure gauge accuracy is within 2 inches.

Proof of concept in 2 to 4 weeks (baseline corridor scans, accuracy validation, CMMS connectivity), pilot in 6 to 8 weeks on priority corridors with work order integration and violation KPIs, then a seasonal monitoring programme with quarterly model updates and drift monitoring. Tunnels are excluded from vegetation monitoring; structure gauge monitoring continues through them.

LiDAR still wins for millimetre clearance certification of new rolling stock. RailGuard is the continuous, lower-cost layer between precision surveys, and most railroads that use it keep both.

Deployment complete in 60 to 90 days. Survey time typically falls 35 to 45 percent (weeks to days per subdivision), 60 to 80 percent more encroachments are found than by manual patrol, and crew dispatch costs come down 40 to 55 percent.