# NADIR Shadow Scorecard

Generated: **2026-06-13 19:02 UTC** · Suite runtime: **4.3s**

> Shadow-mode perception reliability benchmarks — **not** automotive safety certification.
> For LOI and fleet pilot conversations only.

## Aggregate metrics (bootstrap 95% CI)

| Metric | Current | 95% CI | Golden | Baseline tag |
|--------|---------|--------|--------|--------------|
| IDF1-style track continuity | 1.0000 | 1.0000 [1.0000, 1.0000] | 1.0000 | — |
| MTTD (s) | 0.0100 | 0.0100 [0.0000, 0.0300] | 0.0100 | — |
| False alerts / 1000 km-shadow | 0.0000 | 0.0000 [0.0000, 0.0000] | 0.0000 | — |
| Tier NLL | 0.0000 | 0.0000 [0.0000, 0.0000] | 0.0000 | — |

**Regression gate:** PASS (tolerance 2.0%)

## Scenarios

| Scenario | OEM profile | ID continuity | MTTD (s) | False alerts / 1k km | Tier NLL |
|----------|-------------|---------------|----------|----------------------|----------|
| garage_creep | Parking Pilot OEM | 1.0000 | 0.00 | 0.000 | 0.0000 |
| highway_weave | Example Motors | 1.0000 | 0.00 | 0.000 | 0.0000 |
| imu_bias | Example Motors | 1.0000 | 0.00 | 0.000 | 0.0000 |
| parking_creep | Parking Pilot OEM | 1.0000 | 0.00 | 0.000 | 0.0000 |
| sensor_dropout | Example Motors | 1.0000 | 0.00 | 0.000 | 0.0000 |
| straight_highway | Example Motors | 1.0000 | 0.00 | 0.000 | 0.0000 |
| time_sync_slip | Example Motors | 1.0000 | 0.00 | 0.000 | 0.0000 |
| urban_brake | Metro ADAS | 1.0000 | 0.10 | 0.000 | 0.0000 |
| urban_night_rain | Metro ADAS | 1.0000 | 0.00 | 0.000 | 0.0000 |
| windshield_swap_proxy | Metro ADAS | 1.0000 | 0.00 | 0.000 | 0.0000 |

## Stratified by OEM sensor profile


## Scorecard v2 extensions

| Extension | Value |
|-----------|-------|
| Conformal coverage target | 0.95 |
| W₂ shift AUC | 0.0000 |
| Lab parity max tier step | 1 |

### Example Motors

- **IDF1-style track continuity:** 1.0000
- **MTTD (s):** 0.0000
- **False alerts / 1000 km-shadow:** 0.0000
- **Tier NLL:** 0.0000

### Metro ADAS

- **IDF1-style track continuity:** 1.0000
- **MTTD (s):** 0.0333
- **False alerts / 1000 km-shadow:** 0.0000
- **Tier NLL:** 0.0000

### Parking Pilot OEM

- **IDF1-style track continuity:** 1.0000
- **MTTD (s):** 0.0000
- **False alerts / 1000 km-shadow:** 0.0000
- **Tier NLL:** 0.0000

## World-class ladder (internal)

```
Metric                    Current    Golden     Git tag
────────────────────────────────────────────────────────
IDF1-style track continui   1.0000        1.0          —
MTTD (s)                    0.0100       0.01          —
False alerts / 1000 km-sh   0.0000        0.0          —
Tier NLL                    0.0000        0.0          —
```

## Future validation (Commit 11+)

Third-party shadow validation (e.g. TÜV-style technical review of benchmark methodology,
OEM ground-truth datasets, and safety-case boundaries) is **out of scope** for this scorecard.
NADIR Commit 11+ will document an engagement path; this repository remains pilot-ready SDK only.

---

*Run locally:* `make bench-shadow` from the nadir repo root.