AI-Powered Search and Rescue Prediction System
A Monte Carlo simulation platform that predicts the probable location of missing persons based on behavioral psychology, terrain analysis, and environmental factors.
Waypoint uses agent-based simulation informed by ISRID (International Search and Rescue Incident Database) research to generate probability heatmaps showing where a missing person is most likely to be found. The system considers:
- Human Behavior Patterns - Direction traveling, route following, random walking, view enhancing, staying put
- Terrain Analysis - Elevation, slope (Tobler's hiking function), trails, roads, water features
- Environmental Factors - Temperature, precipitation, wind speed
- Personal Profile - Age, gender, experience level
deltahacks-2026/
├── backend/ # Python FastAPI server
│ ├── app/
│ │ ├── simulation/ # Monte Carlo agent simulation
│ │ ├── dem/ # Digital Elevation Model loader
│ │ ├── terrain/ # Terrain analysis pipeline
│ │ └── routes/ # API endpoints
│ └── tests/
└── frontend/ # React + Vite application
├── src/
│ ├── components/ # Heatmap, Loading animations
│ └── pages/ # HomePage, MapPage
└── public/
- Python 3.11+
- Node.js 18+
- Mapbox API key
cd backend
python -m venv venv
source venv/bin/activate # Windows: venv\Scripts\activate
pip install -r requirements.txt
cp .env.example .env # Add your API keys
uvicorn app.main:app --reloadcd frontend
npm install
cp .env.example .env # Add VITE_MAPBOX_TOKEN
npm run dev| Strategy | Probability | Behavior |
|---|---|---|
| Direction Traveling | 55.9% | Maintains fixed heading |
| Route Traveling | 37.7% | Follows trails/roads |
| Random Walking | 5.5% | Random direction |
| View Enhancing | 0.6% | Seeks high ground |
| Staying Put | 0.3% | Stays near last known position |
- Base Speed: 1.317 m/s (male), 1.241 m/s (female)
- Age Decay: -0.012 m/s per decade after 25
- Tobler's Function: Slope-adjusted movement speed
- Weather Penalty: 8% reduction for rain/snow
| Endpoint | Method | Description |
|---|---|---|
/api/v1/search |
POST | Run prediction simulation |
/api/v1/predict |
POST | Detailed time-series prediction |
/api/v1/status |
GET | Server health check |
{
"latitude": 51.1784,
"longitude": -115.5708,
"age": 35,
"gender": "male",
"skill_level": 3,
"time_last_seen": "2024-01-11T06:00:00Z"
}| Technology | Purpose |
|---|---|
| Python 3.11+ | Runs the Monte Carlo agent simulation with 500+ concurrent agents |
| FastAPI | Handles REST endpoints (/search, /predict) with async support for concurrent requests |
| NumPy | Manages 50×50 probability grids and agent position arrays for heatmap generation |
| Rasterio | Parses NASA SRTM GeoTIFF tiles to extract elevation at any lat/lon coordinate |
| SciPy | Applies gaussian_filter (σ=0.5) to smooth raw agent density into viewable heatmaps |
| httpx | Fetches real-time weather from Open-Meteo API and trail data from Overpass API |
| tqdm | Displays simulation progress bar during Monte Carlo timesteps |
| Technology | Purpose |
|---|---|
| React 18 | Renders the search form, time slider, and playback controls as reusable components |
| Vite | Hot-reloads frontend changes in <100ms during development |
| Mapbox GL JS | Displays interactive satellite/terrain map with probability heatmap overlay layer |
| TanStack Query | Caches API responses and handles loading/error states for prediction requests |
| TypeScript | Type-checks API response schemas (ServerGridResponse, GridMetadata) |
| Source | Purpose |
|---|---|
| NASA SRTM DEM | Provides 30m-resolution elevation for Tobler's hiking function slope calculations |
| Open-Meteo API | Returns current temperature, wind speed, and precipitation for weather penalties |
| OpenStreetMap | Supplies trail/road geometries to weight agent movement toward linear features |
Built with ❤️ at DeltaHacks 2026
MIT License