Deep Reinforcement Learning for Autonomous Underwater Navigation: A Comparative Study with DWA and Digital Twin Validation
Sensors · 01.04.2026 · 26 (7) · 2179
DOI: 10.3390/s26072179 ↗Kurz gefasst
Fachartikel in der Zeitschrift Sensors, online erschienen am 01.04.2026. Verfasst von Zamirddine Mari und 2 weiteren Autor:innen. Beteiligte Einrichtung: DGA Techniques Navales, Direction Générale de l'Armement, 83000 Toulon, France. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 17 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Autonomous navigation in underwater environments is challenged by the absence of GPS, degraded visibility, and submerged obstacles. This article investigates these issues using the BlueROV2, an open platform for scientific experimentation. We propose a deep reinforcement learning approach based on the Proximal Policy Optimization (PPO) algorithm, using an observation space that combines target-oriented navigation information, a virtual occupancy grid, and raycasting along the boundaries of the operational area. This information is encoded into a high-dimensional observation space of 84 dimensions, providing the agent with comprehensive local and global situational awareness. The learned policy is compared against a reference deterministic kinematic planner, the Dynamic Window Approach (DWA), a robust baseline for obstacle avoidance. The evaluation is conducted in a realistic simulation environment and complemented by validation on a physical BlueROV2 supervised by a 3D digital twin of the test site, reducing risks associated with real-world experimentation. The results show that the PPO policy consistently outperforms DWA in highly cluttered environments, notably thanks to better local adaptation and reduced collisions. Finally, experiments demonstrate the transferability of the learned behavior from simulation to the real world, confirming the relevance of deep RL for autonomous navigation in underwater robotics.
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- DGA Techniques Navales, Direction Générale de l'Armement, 83000 Toulon, France.
- Laboratoire d'Informatique et des Systèmes, CNRS UMR 7020, Aix-Marseille University, 13013 Marseille, France.
ORCID-Profile: Zamirddine Mari, Mohamad Motasem Nawaf, Pierre Drap
Zitieren
APA
Mari, Z., Nawaf, M. M., & Drap, P. (2026). Deep Reinforcement Learning for Autonomous Underwater Navigation: A Comparative Study with DWA and Digital Twin Validation. Sensors, 26(7), 2179. https://doi.org/10.3390/s26072179
Vancouver
Mari Z, Nawaf MM, Drap P. Deep Reinforcement Learning for Autonomous Underwater Navigation: A Comparative Study with DWA and Digital Twin Validation. Sensors. 2026;26(7):2179. doi:10.3390/s26072179
BibTeX
@article{Mari2026,
author = {Mari, Zamirddine and Nawaf, Mohamad Motasem and Drap, Pierre},
title = {Deep Reinforcement Learning for Autonomous Underwater Navigation: A Comparative Study with DWA and Digital Twin Validation},
journal = {Sensors},
year = {2026},
volume = {26},
number = {7},
pages = {2179},
doi = {10.3390/s26072179}
}
EUVIMED