Authors
- Princecharles C. Anyadiegwu (corresponding, sole author),
Department of Surveying & Geoinformatics, Imo State University, P.M.B. 2000, Owerri, Nigeria. Email: cpcanyadiegwu@gmail.com
Abstract
The integration of artificial intelligence (AI) into hydrographic surveying represents a paradigm shift in how marine environments are mapped, monitored, and managed. This systematic review synthesized 187 peer-reviewed studies published between 2015 and 2026 to examine the current state, emerging trends, and future trajectories of AI applications in bathymetric data acquisition, processing, and interpretation. Our analysis revealed that deep learning architectures, particularly convolutional neural networks (CNNs), U-Net variants, and transformer-based models, have achieved classification accuracies exceeding 94% in seabed characterization and reduced data processing times by up to 88% compared to traditional manual methods.
The global autonomous underwater vehicle (AUV) market is projected to reach USD 4.64 billion by 2030, growing at a compound annual growth rate (CAGR) of 8.2%, driven largely by AI-enhanced navigation and mapping capabilities. Statistical analysis demonstrates significant performance improvements across all surveyed domains (ANOVA: F = 142.3, p < 0.001), with hybrid AI-geostatistical approaches showing the highest accuracy (96.8% ± 1.9%).
Despite these advances, challenges persisted in data annotation quality, computational requirements, model interpretability, and cross-domain generalization. This review identified six critical research gaps and proposes a technology roadmap spanning three phases, automation (2025 to 2028), autonomy (2028 to 2032), and cognition (2032 to 2035), to guide future development. Our findings indicated that AI is not merely augmenting existing hydrographic workflows but fundamentally redefining the possibilities for ocean mapping at unprecedented scales and resolutions.
Keywords:
AI; Hydrographic Surveying; Deep Learning; Bathymetry, CNNs; Seabed Classification