Damage
Curb strikes & minor collisions
Low-speed impacts shift camera and radar mounts without triggering driver warnings — residual scoring catches extrinsic bias early.
Amazon Logistics · ADAS calibration intelligence
NADIR detects camera, radar, and fusion extrinsic drift after collisions, curb strikes, and service events — before AEB under-performance shows up in a bay scan or incident review. This pilot runs in shadow mode: no dispatch changes, no driver alerts, full evidence trail.
Why now
Damage
Low-speed impacts shift camera and radar mounts without triggering driver warnings — residual scoring catches extrinsic bias early.
Thermal
Aluminum mounts expand across depot heat swings. NADIR models thermal drift separately from collision impulse (see Calibration Lab).
Evidence
Signed evidence bundles tie tier transitions to sensor residuals — audit-ready for fleet safety and insurance workflows.
Success metrics
Targets from pilots/amazon-logistics/success_metrics.json — refined at kickoff workshop.
| Metric | Target |
|---|---|
| Drift detection rate (vs service events) | ≥ 85% |
| False positive rate (shadow labels) | ≤ 12% |
| Median time to calibration triage | ≤ 48 h |
| Tier accuracy vs OEM/bay sample | ≥ 80% |
| Evidence bundle completeness | 100% |
Data requirements
POST /v1/telemetry/ingest or secure batch JSONLWe will send the full brief, LOI exhibit A, and a live Calibration Lab walkthrough tailored to ProMaster-class damage scenarios.
Non-binding design-partner materials · Pilot ID amazon-logistics-shadow-2026 ·
Source package