Wiki source code of PSC.02 — Victim Localization Under Uncertainty
Last modified by Mark Rinse van Koningsveld on 2025/09/02 12:06
Show last authors
| author | version | line-number | content |
|---|---|---|---|
| 1 | Problems include smoke, dust, debris, occlusions; clutter. Analysts want synchronized layers and confidence with “why” (explainability). Dogs and technical search remain crucial; drones/robots are additional tools. | ||
| 2 | |||
| 3 | **Symptoms & Impacts** | ||
| 4 | |||
| 5 | * Time wasted on low-yield spots; missed survivable voids. | ||
| 6 | * TLs receive heatmaps instead of actionable waypoints. | ||
| 7 | |||
| 8 | **Contributing Factors** | ||
| 9 | |||
| 10 | * Desynced feeds; weak provenance (who marked what, when); no rapid re-task loop. | ||
| 11 | |||
| 12 | **Desired Outcome** | ||
| 13 | |||
| 14 | * High true-positive rate; marked victim POIs with confidence and rationale; quick re-inspection. | ||
| 15 | |||
| 16 | **How SYNERGISE addresses** | ||
| 17 | |||
| 18 | * **UC02.0**: AI-assisted victim cues; analyst verification; waypoint marking with notes. | ||
| 19 | * **UC04.0**: robot re-tasking to suspected voids; close-in inspection. | ||
| 20 | * Provenance trail on markers; push **decisions & routes**, not raw feeds. | ||
| 21 | |||
| 22 | **Metrics** | ||
| 23 | |||
| 24 | * Confirmed-victim detection rate; time to first confirmed victim; false-positive burden. |