AI-Enabled Underwater Autonomous Robots for Marine Exploration and Environmental Monitoring: A Technical and State-of-the-Art Review

Abstract

The world's oceans encompass over 70% of the Earth's surface, yet more than 80% of the benthic marine environment remains unmapped, unobserved, and unexplored. Traditional marine exploration mechanisms rely on crewed oceanographic vessels, tethered Remotely Operated Vehicles (ROVs), or conventional Autonomous Underwater Vehicles (AUVs) limited by rigid pre-programmed trajectories and high operational costs. The integration of Artificial Intelligence (AI) into underwater robotics has precipitated a transformative paradigm shift, enabling self-directed navigation, dynamic target tracking, real-time sensor fusion, and autonomous environmental monitoring. This review paper presents a comprehensive, journal-level analysis of AI-enabled underwater autonomous robots for deep-sea marine exploration and environmental monitoring. We evaluate the core technical pillars of modern AI-AUVs, including edge-computed deep neural networks (DNNs), acoustic-optical Simultaneous Localization and Mapping (SLAM), reinforcement learning (RL) for hydrodynamic control, and multi-agent swarm intelligence. Furthermore, we synthesize empirical performance metrics regarding navigation drift error reduction, underwater image restoration algorithms (e.g., Water-GAN and physics-informed neural networks), microplastic/plume detection accuracy, and bio-inspired soft robotics. Crucial research gaps—such as severe acoustic bandwidth constraints, extreme optical attenuation, biofouling degradation, and real-time inference energy budgets—are systematically analyzed alongside actionable future directions. The findings provide a definitive blueprint for oceanographers, marine roboticists, and environmental policymakers working toward autonomous ocean sensing networks.

KEYWORDS: Autonomous Underwater Vehicles (AUVs), Marine Robotics,Artificial Intelligence, Edge SLAM, Deep-Sea Bathymetry, Coral ReefMonitoring, Acoustic Communication, Swarm Intelligence.