Long-Term Mapping of the Douro River Plume with Multi-Agent Reinforcement Learning
Published in ICRA 2026 (under review), 2025
We present a scalable multi-agent system for long-term environmental monitoring using autonomous underwater vehicles. By intelligently coordinating robot teams with intermittent communication, Gaussian process regression, and reinforcement learning, we achieve better accuracy and over double the mission endurance compared to traditional approaches.