Atlas builds the geospatial engine that turns raw LiDAR, imagery and telemetry into measured, costed asset condition — sensor to decision, entirely in-house. Running today across highway networks and airfields.
A proprietary survey-grade 3D engine — sensor-agnostic across vehicle-mounted, handheld and aerial LiDAR — turns raw sensor data collected by us or by you into Infrastructure Intelligence your engineers can action.
Per-pixel LiDAR calibration via plane-fitting and Bayesian optimisation, correcting systematic sensor error before measurement.
A full quaternion transform chain from raw sensor polar coordinates through vehicle frame to world ENU, fused with IMU/GPS.
LiDAR 3D, video imagery and GPS/IMU dynamics fused for high-confidence defect identification and measurement.
Performance-critical tile rasterisation compiled to native code with Cython for production throughput at network scale.
PostgreSQL row-level security with account-scoped data isolation — each authority and operator sees only its own network.
A file-based ETL pipeline with dependency resolution, logging and error recovery, processing real survey data daily.
ML semantic segmentation of dense LiDAR clouds — ground, vegetation, structures and infrastructure — with standards-based class taxonomies validated per class.
Quantitative 3D cloud-to-cloud comparison across repeat surveys with confidence-aware screening, plus DTM/DSM terrain products for corroboration.
GeoJSON, Parquet, COG and LAS outputs with versioned schemas, plus map tiles and streamed 3D point clouds explorable in any browser.
The same discipline applies before a single point reaches the pipeline. See how we engineer capture →
One vertically integrated pipeline replaces a fragmented toolchain. Capture is simply stage one: sensors on any vehicle collect the data, and the 25-stage pipeline does the rest.
Vehicle-mounted sensors collect LiDAR point clouds, high-resolution video and GPS/IMU telemetry simultaneously. One drive captures everything needed for comprehensive condition assessment — highway or airfield.
A YOLO-trained neural network identifies every defect — potholes, cracking, surface failures, vegetation encroachment. A vision-language model provides secondary, multi-criteria condition scoring at scale.
Calibrated 3D LiDAR delivers millimetre-precision measurement of every defect — width, depth, length and area. Surface-fitting algorithms extract clean metrics from noisy point clouds. No manual measurement, no estimation.
Each defect is classified by severity tier and assigned a repair cost from configurable cost-per-m² models. Road and runway segments accumulate total repair liability — the entire network gets a price tag.
Defects are clustered into work packages, prioritised by condition risk, and scheduled against fiscal-year budgets. Contractors receive one-click GeoJSON / Shapefile exports for field crews.
Multi-tenant SaaS with interactive maps, defect inspection, automated costing and budget planning — the full workflow, from capture to funded work package.
Open, fixed and unreviewed defects with millimetre measurements, severity and location — across the whole network, automatically populated from each survey.
Area, max depth, width, length, category and estimated repair cost — extracted from calibrated 3D LiDAR for every defect.
Allocate a fiscal-year budget across the network, ranked by condition risk, and see exactly what each pound funds before crews mobilise.
Good intelligence starts with good data. Where standard capture isn't precise enough, or the environment won't allow it, we engineer the sensor platform ourselves — built for the specific constraints of live infrastructure, not a generic rig.
Vehicle-mounted LiDAR, video and GPS/IMU engineered to hold survey-grade calibration at highway speed — on our fleet or yours.
Sensor platforms and survey procedures engineered around airfield access, safety clearance and precision defect intelligence.
The same precision in a portable platform, for footways and areas a vehicle can't reach.
The calibration and coordinate-transform layer that makes this possible is the same one described above — which is why the pipeline reads our rigs or your existing fleet-mounted sensors identically. Engineering the platform doesn't mean requiring it.
The surveying status quo — manual inspections, annual frequency, and video-only "AI" inspections — was built for an era before AI, LiDAR and cloud. The result is reactive maintenance that costs 5–10× more than prevention.
Survey a network at a fraction of SCANNER cost and turn the data into a defensible, risk-based maintenance programme — exactly what the Well-managed Highway Infrastructure Code of Practice demands.
Runways, taxiways and aprons carry zero tolerance for surface failure. Atlas detects and measures airside pavement defects to millimetre precision, so engineering teams can plan interventions around operations.
The market is split between survey specialists who don't manage assets and asset platforms that don't detect defects. Atlas OS is the only system that closes the loop — detection through to funded schedule.
| Atlas OS | Vaisala / SCANNER | Gaist | Yotta | Confirm | |
|---|---|---|---|---|---|
| AI defect detection | YOLO + LLM | Rules-based | Camera AI | Manual | Manual |
| 3D LiDAR measurement | mm precision | Yes | No LiDAR | No | No |
| Automated costing | Per defect | No | No | Manual entry | Template |
| Budget planning | Fiscal year | No | No | Yes | Basic |
| Work scheduling & export | Packages + export | No | No | Yes | Yes |
Live across local government, infrastructure contractors and transport operators — processing real survey data through the full workflow today.
Book a demo and we'll walk you through a live survey — from raw LiDAR to a funded, prioritised maintenance programme for your highways or airfield.