RescueNet is an agentic AI-powered emergency response system designed to improve communication and coordination during medical, disaster, and crime-related emergencies. The system leverages a multi-agent architecture, mesh networking, and real-time prioritization to ensure faster, smarter, and more reliable response, even in degraded connectivity environments.
-
Multi-Modal Input
- Accepts both text and voice (TTS/ASR) input from users in emergencies.
-
Mesh Networking
- Users act as nodes in a mesh network using Bluetooth/Wi-Fi.
- Ensures communication continuity even with degraded infrastructure.
-
Routing Agent
- Detects the type of emergency (Medical, Disaster, Crime).
- Routes input to specialized agents.
-
Specialized Agents
- Medical Agent → integrates a Medical Text Analyzer for health emergencies.
- Disaster Agent → recognizes disaster-related scenarios.
- Crime Agent → handles criminal emergency cases.
- Sentiment Agent → adjusts priority in real-time based on severity and transcription.
- Allocator Agent → allocates resources, finds nearest hospitals, relief centers, or police stations.
-
Real-Time UI for Responders
- Interactive Google Maps with priority tags.
- Navigation support to emergency sites.
- Call-to-action prompts for dispatching resources.
-
Tool Integration
- Google API Tools → Maps + Search for resource allocation.
- Medical Text Analyzer → Identifies symptoms, severity, and urgency.
- Sense → User submits emergency input (text/voice).
- Plan → Routing Agent identifies type + allocates to specialized agents.
- Act → Agents analyze context, allocate resources, and tag priority.
- Observe → Sentiment & Allocator Agents re-check and reprioritize in real-time.
- AI/ML: LLMs for language understanding, sentiment analysis.
- Networking: Bluetooth Mesh (degraded), Wi-Fi (high bandwidth).
- APIs: Google Maps API, Google Search API, Medical Text Analyzer.
- Frontend: Rescue Responder UI (interactive maps, navigation, CTAs).
- Backend: Multi-agent orchestration & routing engine.
- Maintaining connectivity in low-network areas via mesh systems.
- Dynamic prioritization using Sentiment Agent.
- Handling multi-modal inputs (text + voice).
- Balancing resource allocation across hospitals, relief areas, and police stations.
- Working prototype tested on simulated crisis scenarios.
- Multi-agent reasoning loop successfully demonstrated.
- Seamless integration with Google Maps for live response coordination.
- Expand to multilingual emergency handling.
- Integrate drones & IoT devices for faster ground response.
- Deploy at scale for disaster-prone rural regions.
- Sehat gang
- Members: Hamza, Ali, Maaz