Skip to content

About

Multimodal AI platform for disaster response — triages incoming aid requests with Gemini, matches volunteers by skill and proximity, and gives coordinators a real-time map of active incidents.

Topics

Resources

Stars

2 stars

Watchers

0 watching

Forks

Repository files navigation

🛰️ ResourceRadar AI

Real-time Humanitarian Resource Allocation via Multimodal AI

License Next.js Flutter Gemini

ResourceRadar AI is a state-of-the-art humanitarian platform designed to bridge the "last-mile" gap in disaster response. By combining multimodal AI triage with vector-based volunteer matching, it ensures that help reaches those who need it most, in seconds, not hours.


🚀 Key Features

  • 🧠 Multimodal AI Triage: Automated verification of help requests using Google Gemini 1.5 Pro. Analyzes text (English/Hindi/Hinglish) and photos to confirm legitimacy, calculate urgency (0-100), and categorize the incident type.
  • ⚡ Intelligent Volunteer Matching: AI-driven vector semantic matching that automatically routes the closest, best-equipped volunteers to specific crises based on skills and assets, instead of a traditional first-come-first-serve model.
  • 📱 Dual-App Ecosystem (Mobile):
    • Field Agent App: Flutter-based, ultra-low-bandwidth reporting tool for trusted first responders with offline-first data capture support.
    • Volunteer App: Task-focused Flutter interface for responders with real-time mission routing and a global needs map.
  • 🗺️ Coordinator Dashboard (Web): Real-time Next.js command center with live signal feeds, interactive maps, and AI analysis breakdowns for incident commanders.
  • 🤖 Zero-Friction Chatbot Access: Telegram and SMS integrations allowing victims to report emergencies instantly without downloading an app in low-bandwidth zones.
  • 🕵️ Automated Duplicate Detection: Gemini vision models cross-reference incoming photos to automatically cluster redundant reports and flag suspicious claims, preventing wasted resources.
  • 📡 Real-time Event Backend: Firebase Firestore synced with event-driven processors and location intelligence for automatic geocoding and map plotting of critical signals.

🏗️ Architecture

graph TD
    A[Field Agent App] -->|Photo + Text| B(Firestore)
    B -->|Trigger| C[Triage Service]
    C -->|Gemini 1.5 Flash| D{Verification & Urgency}
    D -->|Verified| E[Matching Service]
    E -->|Vertex AI| F[Volunteer App]
    F -->|Accept Mission| G[Mission Routing]
    D -->|Insights| H[Coordinator Dashboard]
Loading

📂 Repository Structure

  • apps/field-agent/: Flutter app for reporting needs.
  • apps/volunteer/: Flutter app for responders.
  • apps/coordinator-nextjs/: Next.js 14 management dashboard.
  • backend/triage-service/: Python/Gemini service for signal analysis.
  • backend/matching-service/: Python service for volunteer allocation.

🛠️ Quick Start

1. Prerequisites

  • Flutter SDK (3.16+)
  • Node.js (18+)
  • Python (3.10+)
  • Google Cloud Project with Gemini API & Maps SDK enabled.

2. Environment Setup

Copy .env.example to .env in the root directory and add your API keys:

GEMINI_API_KEY=your_key
NEXT_PUBLIC_GOOGLE_MAPS_API_KEY=your_key
GOOGLE_CLOUD_PROJECT=your_project_id

3. Run Locally

# Start Triage Service
cd backend/triage-service && pip install -r requirements.txt && python src/listener.py

# Start Coordinator Dashboard
cd apps/coordinator-nextjs && npm install && npm run dev

⚖️ License

Distributed under the Apache License 2.0. See LICENSE for more information.


Built with ❤️ for humanity by [ResourceRadar Team]

About

Multimodal AI platform for disaster response — triages incoming aid requests with Gemini, matches volunteers by skill and proximity, and gives coordinators a real-time map of active incidents.

Topics

Resources

Stars

2 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages