Drones to the Rescue: AI Framework for Disaster Relief
Published:
Overview
An AI-driven framework for deploying autonomous drones equipped with LTE relays to provide emergency network coverage and assist in locating survivors after natural disasters.
Abstract
In the aftermath of natural disasters, rapid restoration of communication networks is crucial for effective relief operations. This paper presents an AI-driven framework for deploying autonomous drones equipped with LTE relays to provide emergency network coverage and assist in locating survivors.
Key Contributions
- Implemented Q-learning model to dynamically determine optimal drone placements based on coverage area
- Developed A* pathfinding algorithm to guide emergency vehicles to target locations while avoiding obstacles
- Achieved 100% area coverage with only 4 drones, outperforming PPO and MCTS baselines
- Verified convergence over 20,000+ training episodes
Results
Simulation results demonstrate that this integrated approach significantly improves coverage area and provides a quick path to the target position. The proposed system offers a scalable solution for intelligent, drone-based disaster response.
Technical Skills
Python, Q-Learning, Reinforcement Learning, A* Pathfinding, Proximal Policy Optimization