SafeRoute 🛡️

AI-Powered Night-Time Safe Navigation Web Platform

"Because the fastest way home is not always the safest."

Built by Team FusionX (Kumari Khushi, Swati Pal) — CodeFusion 2026 · Problem Track: Smart Mobility & Public Safety

---

📌 The Problem

Mainstream map apps optimize for speed and distance — they never factor in street lighting, crowd presence, or crime history. At night, the shortest path often cuts through isolated lanes, unlit stretches, and low-footfall areas.

• 52% of urban Indian women say they always or often feel unsafe walking alone at night (YouGov India Safety Survey)
• Women, students, and night-shift workers disproportionately face this gap every night
• The people most likely to need a safer route have the least help finding one

SafeRoute closes this gap.

💡 The Solution

SafeRoute is a safety-first navigation layer that ranks routes by safety first, distance second — the opposite of conventional map apps.

Conventional Map Apps	SafeRoute
❌ Shortest / fastest path only	✅ Safest, well-lit path prioritized
❌ Traffic-aware, safety-blind	✅ Lighting + crowd + CCTV aware
❌ Ranks purely by distance & ETA	✅ Ranks by community-verified safety
❌ No community safety input	✅ Real-time reports keep data fresh

✨ Enterprise-Grade Data Architecture

We didn't just mock data for a hackathon. Our backend is a genuine, scalable data powerhouse:

• 🛰️ NASA GIBS Live Image Processing: SafeRoute dynamically fetches NASA Earth at Night (VIIRS) satellite imagery tiles for any route and physically scans the pixels using Node.js to calculate macro neighborhood brightness.
• 🗺️ Live OSM Geo-Caching: We query the Overpass API in real-time to fetch lit=no (broken streetlights), amenity=police (CCTV proxy), and crowd-generating POIs (cafes, shops). Results are geo-cached in a 1km grid for 0ms latency during navigation.
• ⏰ Time-of-Day Crowd Heuristics: Our engine calculates crowd density by cross-referencing OSM POI counts with the current time (e.g., a street full of closed shops at 3 AM is penalized as an isolated "ghost town").
• 📊 Nearest-Neighbor Crime Index: We built a curated historical NCRB proxy dataset of 145 Indian cities (5 per state). The engine uses the Haversine formula to detect the user's closest city and pulls its baseline crime risk.
• 🧠 Generative AI Route Summaries: Powered by OpenRouter and Google Gemini 2.0 Flash Lite, delivering natural-language route guidance explaining exactly why a route is safer.
• 📡 Real-Time GPS SOS & Tracking: Built-in hardware-accelerated GPS polling (watchPosition) connected via WebSockets. No mock data—live location is securely broadcasted directly to emergency contacts.
• 🚨 Twilio-Integrated SOS System: A single tap immediately triggers an SMS broadcast to predefined emergency contacts with a dynamically generated live-tracking URL and current coordinates.
• 🔐 Stateless Security: Fully authenticated using bcrypt password hashing and secure JSON Web Tokens (JWT) for session management, ensuring robust security in production.

🧰 Tech Stack

Layer	Technology
Frontend	React (Vite) + Tailwind CSS + React-Leaflet
Backend	Node.js + Express + jimp (NASA Image Processing)
Real-time	Socket.IO (WebSockets)
Database	MongoDB (User data + Live Community Hazards)
Security	JWT (Stateless Sessions) + bcrypt
Routing	OSRM (Open Source Routing Machine)
Safety Engine	Node.js Multi-Layer Evaluator (NASA + OSM + NCRB + MongoDB)
AI	OpenRouter (Gemini 2.0 Flash Lite)
Communications	Twilio SMS API

📂 Project Structure

saferoute/
├── client/                  # React frontend
│   └── src/
│       ├── components/      # map, report, sos, search, layout
├── server/                  # Node/Express backend
│   └── src/
│       ├── routes/
│       ├── controllers/
│       ├── services/        
│       │   ├── safetyScoreEngine.js  # The Multi-Layer Math Evaluator
│       │   ├── osmService.js         # Live Overpass Geo-Caching
│       │   ├── nasaService.js        # NASA Satellite Image Processing
│       │   ├── crimeDataService.js   # Nearest-Neighbor historical search
│       │   └── llmService.js         # OpenRouter Gemini Summaries
│       ├── models/
│       └── data/            # 145-city historicalCrimeData.json
└── README.md

🚀 Getting Started

Prerequisites

• Node.js ≥ 18
• MongoDB Atlas connection string (or Local Mongo)
• OpenRouter API Key
• Twilio Account SID, Auth Token, and Phone Number (for SOS SMS)

Installation

# Clone the repo
git clone https://github.com/<your-org>/saferoute.git
cd saferoute

# Install server dependencies
cd server
npm install
# Add OPENROUTER_API_KEY and MONGODB_URI to .env

# Install client dependencies
cd ../client
npm install

Running Locally (Development)

# Terminal 1 — start backend
cd server
npm run dev

# Terminal 2 — start frontend
cd client
npm run dev

App runs at http://localhost:5173, API at http://localhost:5000.

Production Deployment (Monolithic)

SafeRoute is architected for simple, cost-effective single-instance deployment (e.g., Render, DigitalOcean, Railway). The Express backend automatically serves the statically built React frontend.

# 1. Build the frontend
cd client
npm run build

# 2. Start the production server
cd ../server
npm start

The entire application (Frontend + API + WebSockets) will now run securely on port 5000 (or process.env.PORT).

🗺️ Roadmap

Phase	Goal	Deliverable
Phase 1 (Done)	Dynamic Safety-score engine on live planetary data	Working Demo
Phase 2 (Done)	Pilot with a college campus / local ward	Admin Campus Pilot Dashboard Built
Phase 3	Integrate municipal CCTV & smart-city feeds	City-Wide Rollout

🌍 Why It Matters

SafeRoute is built for women, students, and night-shift workers who need to move after dark. It reduces the daily mental calculation of "is taking this route worth the risk?" — and becomes a living, community-built safety layer that gets smarter the more people use and contribute to it.

---

SafeRoute turns navigation from fastest to safest — powered by AI, NASA, OpenStreetMap, and community intelligence.