A real-time AI-powered crisis management system that transforms chaos into coordinated emergency response.
- Team Name: Anavrin
- Team Leader: Sujal Barwad
- Built For: Google Solution Challenge (via Hack2Skill)
Design a robust system to instantly detect, report, and synchronize crisis response across a decentralized hospitality ecosystem.
Current systems suffer from:
- ❌ Fragmented communication
- ❌ No real-time coordination
- ❌ Lack of intelligent evacuation guidance
SENTINEL is a high-stakes emergency management platform designed for complex, multi-floor structures where every second counts. It converts emergencies into a structured, real-time response loop by connecting guests in distress, emergency responders, and an AI-driven intelligence system.
- 1️⃣ Detection: Picks up a crisis from IoT sensors (smoke/fire detectors), Voice reports via a Dialogflow AI agent, or User SOS requests via the PWA.
- 2️⃣ Intelligence (The AI Brain): Once an incident hits the database, Gemini 2.0 Flash analyzes the full floor plan, classifies severity (P0/P1/P2), identifies safe vs. blocked exits, and uses Dijkstra’s Algorithm for optimal routing.
- 3️⃣ Response: Instantly pushes calculated paths to the frontend, blasts Push alerts via Firebase Cloud Messaging (FCM), and provides Multilingual voice guidance (TTS).
- 🔥 Real-time Detection: Immediate response to IoT fire/smoke triggers.
- 🗺️ Dynamic Evacuation Routing: Dijkstra's algorithm computes the safest path live as fires spread.
- 🧠 AI-Generated Intelligence: Gemini generates a Situation Report (SitRep) in milliseconds.
- 🌍 Multilingual Alerts: Instant Text-to-Speech (TTS) announcements in 4 languages (EN, HI, FR, ES).
- 🆘 One-Tap SOS: Guests can trigger direct rescue requests.
- ♿ Mobility-Impaired Detection: AI automatically flags mobility-impaired guests during an evacuation.
- 🧑🚒 Tactical Command Center: Admin dashboard styled like a professional military HUD, not a standard website.
- 🌐 Offline-Capable: Aggressive Service Worker caching ensures the app works during network drops.
- 🏢 Interactive 3D Building Model: Real-time facility visualization using Three.js / React Three Fiber.
✅ The Only Solution combining IoT + Voice AI + Gemini recommended procedures + live evacuation routing in a single system.
✅ Zero-Cost Deployment using the Google Cloud free-tier, removing the #1 barrier for small/mid-size properties.
✅ Mobility-Aware Planning baked into the AI output—no other hotel safety tool does this automatically.
✅ Multilingual Voice Alerts to serve international guests that competitors completely ignore.
| Traditional Systems | SENTINEL |
|---|---|
| Only alarms 🔔 | Full AI situational awareness 🧠 |
| No evacuation guidance | Smart routing 🗺️ |
| Fragmented systems | Unified real-time pipeline 🔗 |
| No multilingual support | Multi-language voice alerts 🌍 |
🎨 Frontend
- React + Vite: PWA dashboard for admin & guests.
- Three.js / React Three Fiber: Interactive 3D routing & satellite mapping.
- Service Worker: Offline caching for crisis resilience.
🧠 AI & Voice
- Gemini 2.0 Flash: Situational intelligence & SitRep generation.
- Dialogflow CX: Voice-to-incident NLP (extracts floor/zone/room).
- Web Speech API: Zero-latency multilingual audio announcements.
⚙️ Backend & Infrastructure
- Python Flask & Google Cloud Run: Event processor & REST API.
- Google Cloud Pub/Sub: Real-time event bus for sensor data.
- Cloud Functions Gen2: Gemini triggers + FCM dispatcher.
- Firebase Ecosystem: Firestore (Real-time NoSQL), Firebase Auth, Firebase Hosting, Firebase Cloud Messaging.
SENTINEL/
├── backend/ # Serverless Python backend
│ ├── cloud-functions/ # Firebase Gen 2 Cloud Functions
│ │ ├── fcm-dispatcher/ # Sends Push Notifications via FCM
│ │ └── gemini-trigger/ # Generates SitRep via Gemini 2.0 Flash
│ ├── event-processor/ # Flask API on Cloud Run (Pub/Sub ingest)
│ └── iot-simulator/ # Simulates IoT fire/smoke sensors
├── docs/ # Project documentation & Architecture
├── public/ # Static assets (3D models, floor plans)
├── src/ # React Frontend Source Code
│ ├── components/ # UI elements (Maps, SitRep, Tables)
│ ├── context/ # React Context (Auth)
│ ├── data/ # Mock JSON data
│ ├── pages/ # Main views (Home, Login, Landing)
│ ├── utils/ # Dijkstra routing and Graph logic
│ ├── App.jsx # App routing and FCM logic
│ ├── firebase.js # Firebase init & offline caching
│ └── main.jsx # React entry point & PWA Service Worker
├── DEPLOYMENT_GUIDE.md # Setup & Deployment instructions
├── firestore.rules # Security rules for Firebase Database
└── package.json # Frontend dependencies
- Advanced AI Reasoning: Integration with Gemini 2.0 Pro for multi-incident correlation to distinguish between isolated glitches and structural threats.
- Auto-Reporting: Instant, AI-generated post-incident summaries for safety audits and compliance teams.
- V2V Incident Debriefing: Allow admins to verbally query historical logs (e.g., "Summarize the East Wing fire") for hands-free analysis.
- Omni-Channel Accessibility: Full Flutter mobile app integration with 10+ language support via Google Cloud Translation.
- 🔗 GitHub Repository: https://github.com/Sriyasnehasis/SENTINEL
- 🎥 Demo Video: Watch on YouTube
- 🎨 MVP (Figma Design): View Wireframes
- 🌐 Working Prototype: Live Application
❤️ Final Note: SENTINEL is built for worst-case scenarios, where every second matters. It aims to save lives using AI, real-time systems, and intelligent coordination.