harshaleemalu03/disaster-management

<div align="center">

🚨 Disaster Response Coordination System

OpenEnv v2.1 — Hackathon Submission

High-fidelity AI environment for emergency dispatch coordination Real travel delays · Fire cascade spread · Fairness-aware dispatch · Live visualization

</div>

🌍 Real-World Motivation

Every 60 seconds of delayed emergency response increases mortality risk. Real dispatch centers must simultaneously manage:

• —Dozens of concurrent incidents across urban, suburban, and rural areas
• —Heterogeneous fleets — ambulances, fire trucks, rescue teams, helicopters — with different speeds and specializations
• —Geographic constraints — urban traffic congestion, rural road conditions, cross-terrain helicopter routes
• —Equity obligations — rural communities must not be systematically deprioritized

Modern CAD systems fail under surge conditions. This environment benchmarks AI agents on these failure cases.

🏗️ Architecture

FastAPI Server (port 7860)
        │
DisasterResponseEnv
        │
 Task1 | Task2 | Task3
        │
   WorldEngine

🎯 Three Tasks

Task 1 — Incident Prioritization [EASY]

• —Rank incidents by urgency
• —Baseline: 1.00

Task 2 — Resource Allocation [MEDIUM]

• —Assign resources with travel delays
• —Baseline: 0.61

Task 3 — Dynamic Coordination [HARD]

• —Full simulation with cascading events
• —Baseline: 0.27

🔥 Key Features

• —🔥 Fire cascade spread
• —🚑 Real distance-based travel
• —⚖️ Fairness-aware dispatch (Gini coefficient)
• —📊 Dense reward system
• —🌐 Live visualization dashboard

🚀 Quick Start

pip install -r requirements.txt
uvicorn server:app --host 0.0.0.0 --port 7860
python inference.py

📺 Endpoints

• —/ → Landing page
• —/demo → Live simulation
• —/docs → API docs
• —/view/{session_id} → Dashboard
• —/visualize/{session_id} → Text grid

📁 Structure

disaster_response/
├── env/
├── server.py
├── inference.py
├── requirements.txt
├── Dockerfile
└── README.md

⚡ Performance

• —< 50 ms / step
• —< 3 sec full run
• —Deterministic & reproducible