RescueNet

Inspiration

Natural disasters such as earthquakes, floods, and wildfires continue to cause massive loss of life due to slow coordination and communication failures during emergency response. Our inspiration came from real-world disaster response limitations, as well as recent tragedies in our area which suffered a huge flood because of an atmospheric river.

We asked a simple question: What if emergency response systems could think, adapt, and coordinate autonomously in real time—even when parts of the system fail?

That question led to RescueNet.

What RescueNet Does RescueNet is a decentralized, multi-agent AI system designed to improve disaster response efficiency and resilience.

It coordinates: Autonomous drones Rescue robots Emergency vehicles IoT environmental sensors Using real-time communication and machine learning, RescueNet dynamically assigns tasks, reallocates resources, and adapts to changing disaster conditions—even if individual agents or infrastructure fail.

How We Built It

RescueNet is built on a multi-agent architecture where each agent operates autonomously while collaborating through asynchronous communication.

Core System Components:

Agent Ecosystem: Drones, vehicles, robots, and IoT sensors act as independent agents Communication Protocol: Mesh networking with priority-based messaging for resilience AI Coordination Layer: Reinforcement learning and supervised models optimize task allocation and deployment Model Architecture: Combines DQN, PPO, and CNNs with attention mechanisms and experience replay Resilience Design: Self-healing agents and decentralized decision-making ensure system continuity The system continuously learns from real-time feedback, allowing it to adapt to different disaster scenarios such as floods, earthquakes, and wildfires.

Challenges We Ran Into

One of the biggest challenges was integrating multiple autonomous systems while maintaining reliable communication and low latency. Coordinating agents that operate independently—yet must collaborate—required careful system design.

We also faced difficulties integrating diverse APIs and ensuring smooth data flow across components, which initially slowed development and testing.

How We Overcame Them

To solve these challenges, we:

Implemented asynchronous, priority-based communication

Designed self-healing agents that detect failures and reassign tasks automatically Used decentralized architecture to eliminate single points of failure Optimized AI coordination through reinforcement learning and adaptive feedback loops These solutions significantly improved system robustness, scalability, and response speed.

Accomplishments We're Proud Of

Built a fully decentralized, fault-tolerant multi-agent system Implemented real-time AI-driven task allocation Designed a system that remains functional even during partial infrastructure failure Created a scalable framework adaptable to multiple disaster types What We Learned Through building RescueNet, we learned that effective autonomy requires more than intelligent models—it requires resilience, adaptability, and human-centered design. We gained hands-on experience with multi-agent systems, reinforcement learning, distributed communication, and large-scale system architecture.

What's Next for RescueNet

Real-world disaster simulation testing Edge computing deployment for faster local decision-making Integration with government and NGO emergency response teams AI-enhanced emotional intelligence for rescue bots to assist survivors psychologically RescueNet is not just a project—it’s a step toward faster, smarter, and more resilient disaster response systems.

Built With

• aceternity
• agents
• cnn
• eslint
• esri
• google-maps
• leaflet.js
• machine-learning
• mesnetworking
• mutli-agent
• react
• reactrouter
• shadcnui
• tailwind
• typescript
• vercel
• vite