Exploring the Combined use of Deep Learning and LiDAR Bathymetry Point-Clouds to Enhance Safe Navigability for Maritime Transportation: A Systematic Literature Review

Maritime transport significantly contributes to global economic growth. The majority of global commerce is transported via sea. Ensuring navigable conditions for maritime transportation is essential to keep ships moving safely and efficiently. Remote sensing technologies, particularly Aerial Laser Scanning (ALS) technology, can be adapted to monitor sub-sea conditions. Based on point clouds from LiDAR Bathymetry, the ALS supports practitioners in classifying submerged objects. However, this technology provides an extensive amount of Big Data which needs to be processed by practitioners. With recent advancements in Deep Learning (DL), multiple researchers have begun exploring DL algorithms to classify Airborne Bathymetry LiDAR point cloud data. This paper employs a Systematic Lit-erature Review (SLR) methodology to examine the applicability of DL in the classification of airborne Bathymetry LiDAR point cloud data. Through the SLR process, we identified nineteen hundred relevant articles from the ScienceDirect and Google Scholar databases over the last decade. A comprehensive analysis of identified articles revealed that the majority employed deep learning models are for classification, detection, and semantic segmentation. The most commonly used algorithms are Con-volutional Neural Networks (CNNs), Multi-Layer Perceptrons (MLPs), Random Forest, and Support Vector Machines. In addition, in the identified articles, it was observed that these algorithms were applied to both waveform data and image data.