Sensor Stack¶
The LAIRA Rig is a calibrated multi-modal sensor package that mounts on existing airport or fleet operations vehicles. This page is the single source of truth for the sensor specifications cited by the compliance matrices. Every row in a matrix that references a measurement resolution traces back to a spec on this page.
Specifications at a glance¶
| Component | Specification | Notes |
|---|---|---|
| LiDAR | 128-channel; sliding-window SLAM supports 32-/64-channel rig variants without point-cloud-density loss | The primary sensor for elevation, depth, and texture measurements |
| RGB Camera | 4K resolution; 1.67 mm/pix at 3 m depth | Primary sensor for distress classification, marking color and pattern, FOD material |
| RTK GPS / GNSS | ±2 cm positional accuracy (WGS84) | Below the sample-unit-boundary tolerance of all federal pavement standards |
| IMU | 6-axis inertial measurement | Vehicle-motion compensation; pairs with visual-inertial odometry for GPS-degraded segments |
| Edge Compute | NVIDIA-accelerated; < 200 ms inference latency | Real-time distress classification during movement |
| Connectivity | 5G/LTE dual SIM + WiFi | Streams to the Airwai Intelligence Dashboard; supports cellular uplink for live monitoring |
| Operating Speed | Up to 55 mph (90 km/h) | Posted-speed inspection on Interstate / NHS without lane closure |
| Data Capture Width | ~55 m | Full lane plus shoulder plus sidewalk in a single pass |
| Weight | ~35 lbs (15.9 kg) | Man-packable; mounts on standard operations vehicles without intensive retrofit |
| Battery | 7 hrs operating / 14 hrs passthrough | Full-shift operations without recharging |
| Calibration | LiDAR + RGB alignment calibration protocol | Calibration certificates managed on a 6-month cadence aligned to AC 150/5300-17C requirements |
Measurement resolution¶
These are the resolution claims the matrices repeatedly cite. Every claim is grounded to a specific component above.
| Measurement | Resolution | Underlying component |
|---|---|---|
| Crack width | ±0.7 mm | RGB camera at 1.67 mm/pix, with sub-pixel refinement |
| Crack length | ±0.7 mm in plane; ±RTK accuracy across segments | RGB + RTK |
| Elevation differential (faulting, depression, drop-off) | ±3 mm | LiDAR DEM |
| Rut depth (transverse profile) | ±3 mm | LiDAR transverse profile |
| Positional accuracy (per-detection geotag) | ±2 cm WGS84 | RTK GNSS |
| Color compliance (marking classification) | Categorical per ML model | RGB color classification |
Reading the resolution claims
When a matrix row says "✅+ Exceeds" against a measurement threshold from a regulatory standard, the comparison is between the relevant resolution above and the standard's smallest tabulated severity threshold. For example, D6433 §X2.7 Table X2.2 tabulates the smallest faulting severity at >⅛ in (≈ 3.2 mm); LAIRA's ±3 mm LiDAR elevation differential sits below that threshold, hence ✅+.
Why each sensor is on the rig¶
LiDAR¶
LiDAR provides the depth and elevation primitives that no RGB-only system can. Every elevation-differential distress in ASTM D5340 (airfield) and ASTM D6433 (roadway) — depression, faulting, rutting, drop-off, swell, shoving, corner spalling, joint spalling — is measured from LiDAR returns. The 128-channel native configuration delivers point-cloud density sufficient for sub-millimeter elevation resolution at highway speeds. The sliding-window SLAM algorithm preserves that density when running 32- or 64-channel cost-down variants, which makes the same algorithmic substrate viable across the rig tier and the lower-cost Vision Lite and Capture tiers.
RGB camera¶
The 4K RGB camera handles distress classification (the "what is this crack and how severe is it" question), marking color compliance, and FOD material identification. At 3 m sensor-to-pavement depth, 1.67 mm/pixel resolution renders cracks 10× narrower than the smallest threshold called out in any pavement standard, so width-based severity classification is bound by sensor calibration, not pixel limits.
RTK GPS / GNSS¶
A ±2 cm WGS84 position fix on every detection is what makes the data set defensible. It allows:
- Per-detection geotag for the audit trail referenced in every matrix
- Linear-Referencing-System (LRS) alignment for state-DOT data submission (Caltrans PaveM, TxDOT PMIS)
- Sample-unit-boundary assignment per D6433 §2.1.7 sample-unit definitions
- Pass-over-pass change detection at sub-cm registration
IMU and visual-inertial odometry¶
The IMU compensates for vehicle motion so the per-frame inference is anchored to a stable platform reference. When GPS signal degrades (under canopy, urban canyon, gantry shadow), visual-inertial odometry maintains positional continuity through the degraded segment. Together they let LAIRA hold the ±2 cm geotag claim through real-world driving conditions.
Edge compute¶
Inference happens on the rig, not in the cloud. The < 200 ms latency budget means real-time alerts (FOD on a movement area, high-severity distress on a hot spot) can fire during the inspection pass, not days later after cloud post-processing. This is the architectural distinction from ARAN-class systems that capture in the field and process in batches.
Host vehicle requirements¶
LAIRA is built to mount on existing operations vehicles already authorized for the inspection area. The rig itself does not introduce new host-vehicle requirements beyond what is already true of any inspection vehicle on the relevant pavement.
- Airfield: AC 150/5210-20A — high-visibility daytime marking, lighted for night ops, rotating beacon always on, two-way radio for ATCT contact.
- DoD installation: AFI 13-204v3 host-vehicle authorization.
- Roadway: state-DOT inspection-vehicle norms; no special permitting beyond standard road-survey conventions.
Mounting hardware is standard tow-hitch or roof-mount kit; no intensive retrofit, no permanent fixture, no host-vehicle return-to-service certification required.
Calibration cadence¶
| Activity | Frequency | Reference |
|---|---|---|
| LiDAR-to-RGB alignment | Semi-annual or pre-deployment | AC 150/5300-17C cites the 6-month cadence for mobile LiDAR |
| Positional accuracy verification (RTK) | Pre-deployment per site | Verified against a known-survey-control point at the site |
| FOD test-object validation | Annual, against AC 150/5220-24 standard 10-item group | DoD installations may require additional cadence per local CE protocol |
Where matrices cite this page¶
| Matrix | Rows that anchor here |
|---|---|
| Civil Airfields | Crack measurement, geolocation, inspection speed, RGB resolution, DEM resolution, sensor calibration |
| Military Airfields | §7 Sensor System Specifications |
| Federal & State Roads | §6 Sensor + Data Output (Roadway Profile) |
| Sidewalks & Curb Ramps | LiDAR DEM elevation primitives (±3 mm); slope resolution (±0.1%) |