Technology · Sensor fabric

How NADIR reads the vehicle sensor topology

ADAS fleets expose cameras, radar, LiDAR, and CAN/IMU streams into one ingest fabric. NADIR scores cross-modal residuals in shadow mode — tiering drift for operators without replacing OEM safety stacks. Scroll through each modality or read every section; the diagram stays visible without JavaScript.

Stage 1 · Camera

Windshield and surround extrinsics

Forward and surround cameras anchor lane geometry, object classification, and many fusion paths. Glass replacement, roof racks, and ride-height changes shift yaw and pitch — often without dashboard faults.

• Residual monitoring after post-repair validation.
• Glass and thermal scenarios in the Calibration Lab.
• Modality-tagged CAUTION tickets for bay scheduling.

Stage 2 · Radar

Corner and front radar boresight

Radar supplies range-rate and all-weather returns. Bumper repairs and bracket tolerances change azimuth; fusion may down-weight echoes silently while ACC still appears nominal to drivers.

• Coupled residuals with camera bearing checks.
• Collision-repair drift patterns in demo scenarios.
• CRITICAL holds before long-haul dispatch.

Stage 3 · Fusion

Cross-modal agreement at the NADIR core

Fusion health is agreement across modalities — not peak signal on one bus. NADIR ingests normalized drift vectors, applies changepoint detection, and publishes NOMINAL, CAUTION, and CRITICAL tiers for operators.

• Shadow scoring parallel to OEM stacks during pilots.
• Evidence bundles on tier transitions.
• API and webhook delivery documented in OpenAPI.

Stage 4 · Fleet queue

From residuals to prioritized interventions

Fleet operations consume tier summaries — ranked VIN queues, closure tracking, and export presets for legal and insurer workflows. Repair networks use the same tiers for franchise scorecards.