Efficient Multi-UAV Path Planning for Wireless Data Harvesting: A Deep Reinforcement Learning Approach

Authors:
Publisher website
Google Scholar
Abstract
This paper presents a multi-agent reinforcement learning (MARL) framework designed to optimize the path planning of multiple unmanned aerial vehicles (UAVs) for wireless data harvesting missions. The proposed approach leverages Double Deep Q-Networks (DDQN) with State-Action-Reward-State-Action (SARSA) and Q-learning algorithms to improve learning efficiency and overall performance.In our framework, UAVs operate in a decentralized manner, learning their policies independently through interactions with a grid-based environment that includes no-fly zones, buildings, start and landing zones, and sensor locations. Extensive simulations demonstrate that the DDQN-based algorithm significantly outperforms SARSA and Q-learning in terms of maximizing data collection, minimizing collisions, and achieving faster convergence, underscoring its potential to enhance UAV operations in dynamic

This publication has 9 references indexed in Scilit: