🚨 RoadSOS — Emergency Response System

Offline-capable, globally applicable emergency response platform for road accident victims. Instantly locates the nearest trauma hospitals, ambulance services, police stations, pharmacies, blood banks, puncture shops, and towing assistance — for any location on earth.

Built for the CoERS IIT Madras Road Safety Hackathon 2026 — Problem Statement 3: Emergency Response Optimization.

Teammates: Tejaswi K - B.Tech AI&DS Poornaa Shree Praveenraj - B.E CSE

---

What RoadSOS Does

When a road accident occurs, every second counts. RoadSOS gives the victim or bystander immediate access to:

• Nearest trauma-capable hospitals — ranked by distance AND capability score, not just proximity
• Ambulance services — both government (108) and private operators
• Police stations and fire stations
• Pharmacies and blood banks
• Puncture shops and towing/garage services
• Country-specific emergency numbers — auto-detected for 200+ countries

The system works offline after first sync — making it viable in rural areas and highway stretches where internet is unavailable.

---

Key Features

• Offline-first architecture — syncs regional data from OpenStreetMap once, stores in local MySQL, runs without internet at query time
• Global applicability — same codebase works for any coordinate on earth; tested across India, Japan, UK, Brazil, UAE, and Kenya
• Capability scoring — hospitals ranked by bed count, specialties, operator type, and emergency facilities — not just distance
• Trauma filter — automatically excludes eye hospitals, dental clinics, maternity homes, and psychiatric facilities from emergency results
• ETA estimation — road-factor distance correction (×1.3) with average urban ambulance speed
• Smart fallbacks — police/fire stations show emergency numbers when direct phone is unavailable; puncture shops fall back to nearest garage
• Country-aware emergency numbers — verified override table for 30+ countries with known dataset errors

---

Tech Stack

Layer	Technology
Backend	Python, FastAPI
Database	MySQL
Map Data	OpenStreetMap via Overpass API
Distance Calculation	Haversine formula (no routing API)
Reverse Geocoding	reverse_geocoder (offline)
Emergency Numbers	Bundled JSON — 200+ countries
Frontend	Streamlit, Folium/Leaflet
Location Resolution	Geopy / Nominatim

---

Project Structure

IITM hackathon/
├── roadsos_backend.py        # Core backend — all logic here
├── emergency_numbers.json    # Bundled emergency numbers — 200+ countries
├── app.py                    # Streamlit homepage
├── styles.css                # Dark theme CSS
├── components/
│   └── header.py
├── pages/
│   ├── 1_Dashboard.py        # About / purpose page
│   ├── 2_Emergency_Services.py
│   ├── 3_Map_View.py
│   ├── 4_SOS.py              # Primary SOS trigger page
│   └── 5_Emergency_Numbers.py
└── services/
    └── api.py                # Backend wrapper for frontend

---

Setup Instructions

Prerequisites

• Python 3.10+
• MySQL 8.0+ running locally
• Internet connection for first regional sync

1. Clone the repository

git clone https://github.com/Whostejaswi/roadsos.git
cd roadsos

2. Install dependencies

pip install streamlit streamlit-folium folium mysql-connector-python \
            reverse_geocoder requests geopy fastapi uvicorn

3. Set up MySQL database

Open MySQL Workbench or MySQL shell and run:

CREATE DATABASE roadsos;
USE roadsos;

CREATE TABLE hospitals (
    id INT AUTO_INCREMENT PRIMARY KEY,
    osm_id VARCHAR(50) UNIQUE,
    name VARCHAR(255),
    lat DOUBLE,
    lon DOUBLE,
    phone VARCHAR(100),
    emergency BOOLEAN DEFAULT FALSE,
    city VARCHAR(100),
    state VARCHAR(100),
    country VARCHAR(10) DEFAULT 'IN',
    confidence ENUM('high', 'medium') DEFAULT 'medium',
    specialty VARCHAR(255),
    capability_score INT DEFAULT 0
);

CREATE TABLE nearby_services (
    id INT AUTO_INCREMENT PRIMARY KEY,
    osm_id VARCHAR(50) UNIQUE,
    name VARCHAR(255),
    service_type VARCHAR(50),
    lat DOUBLE,
    lon DOUBLE,
    phone VARCHAR(100),
    city VARCHAR(100),
    state VARCHAR(100),
    country VARCHAR(10) DEFAULT 'IN'
);

CREATE TABLE regions_synced (
    id INT AUTO_INCREMENT PRIMARY KEY,
    lat DOUBLE,
    lon DOUBLE,
    radius_km INT,
    country_code VARCHAR(10),
    synced_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP
);

CREATE TABLE ambulance_services (
    id INT AUTO_INCREMENT PRIMARY KEY,
    name VARCHAR(255),
    phone VARCHAR(100),
    type ENUM('government', 'private') DEFAULT 'private',
    city VARCHAR(100),
    state VARCHAR(100),
    country VARCHAR(10) DEFAULT 'IN',
    lat DOUBLE,
    lon DOUBLE
);

-- Seed government ambulance data
INSERT INTO ambulance_services (name, phone, type, city, state, country, lat, lon) VALUES
('108 Government Ambulance',    '108',          'government', 'Coimbatore', 'Tamil Nadu', 'IN', 11.0168, 76.9558),
('112 Emergency Response',      '112',          'government', 'Coimbatore', 'Tamil Nadu', 'IN', 11.0168, 76.9558),
('Ziqitza Ambulance (Private)', '1800-313-1414','private',    'Coimbatore', 'Tamil Nadu', 'IN', 11.0168, 76.9558),
('Falck Ambulance',             '044-28282828', 'private',    'Coimbatore', 'Tamil Nadu', 'IN', 11.0168, 76.9558),
('Red Cross Ambulance',         '1800-180-1234','private',    'Coimbatore', 'Tamil Nadu', 'IN', 11.0168, 76.9558);

4. Configure database credentials

Open roadsos_backend.py and update:

DB_CONFIG = {
    "host":     "localhost",
    "user":     "root",
    "password": "your_password",  # ← change this
    "database": "roadsos"
}

5. Run the app

cd "RoadSOS"
streamlit run app.py

The app opens at http://localhost:8501

On first use for any location, the system fetches and caches regional data from OpenStreetMap (takes 30–60 seconds). All subsequent queries run instantly from the local database.

---

How Offline Works

ONLINE (first sync per region)        OFFLINE (all subsequent use)
──────────────────────────────        ────────────────────────────
Overpass API → hospitals table        Haversine query → MySQL
Overpass API → nearby_services        Emergency numbers → local JSON
Mark region as synced in MySQL        Country code → reverse_geocoder

The regions_synced table tracks which areas have been cached. When a new location is detected within 40km of an already-synced region, no internet fetch is triggered.

---

Testing

Run the backend directly to test across multiple locations:

python roadsos_backend.py

This runs the built-in test suite across Coimbatore, Tokyo, London, Krishnagiri, Dubai, São Paulo, and Nairobi — verifying hospitals, services, emergency numbers, and offline fallback behavior for each.

---

API (FastAPI)

Start the REST API separately if needed:

uvicorn main:app --reload --port 8000

Endpoints:

Endpoint	Method	Description
/sos?lat=11.0168&lon=76.9558	GET	Full SOS response for coordinates
/sync?lat=11.0168&lon=76.9558	GET	Trigger regional sync
/health	GET	API health check
/docs	GET	Auto-generated Swagger UI

---

Evaluation Criteria Coverage

Criterion	How RoadSOS addresses it
Reliability and data accuracy	Real OSM data, verified emergency numbers, trauma-capable filter
Number of contacts fetched	Hospitals, police, fire, pharmacy, blood bank, puncture shop, towing
Offline functionality	MySQL cache, local JSON, offline reverse geocoding
Innovation and additional features	Capability scoring, ETA, global applicability, smart fallbacks
Information integration across countries	Tested across 7 countries, auto country detection