A Quasi-Intelligent Maritime Route Extraction from AIS Data

Abstract

The rapid development and adoption of automatic identification systems as surveillance tools have resulted in the widespread application of data analysis technology in maritime surveillance and route planning. Traditional, manual, experience-based route planning has been widely used owing to its simplicity. However, the method is heavily dependent on officer experience and is time-consuming. This study aims to extract shipping routes using unsupervised machine-learning algorithms. The proposed three-step approach: maneuvering point detection, waypoint discovery, and traffic network construction was used to construct a maritime traffic network from historical AIS data, which quantitatively reflects ship characteristics by ship length and ship type, and can be used for route planning. When the constructed maritime traffic network was compared to the macroscopic ship traffic flow, the Symmetrized Segment-Path Distance (SSPD) metric returned lower values, indicating that the constructed traffic network closely resembles the routes ships transit. The result indicates that the proposed approach is effective in extracting a route from the maritime traffic network.

Multi-Density Adaptive Trajectory Clustering Algorithm for Ships Based on AIS Data

“…In terms of ship trajectory prediction [11][12][13], Suo et al [14] extracted a series of trajectories from AIS ship data (i.e., latitude, longitude, speed, and heading), derived the main trajectories by applying the DBSCAN algorithm and finally introduced a deep learning framework and Gate Recirculation Unit model for predicting ship trajectories. In terms of ship route planning [15][16][17], Zhang et al [18] proposed a shortest-path planning method based on AIS data, establishing a low-precision environment model, and determining the grid area of the shortest path through an ant colony algorithm to reduce the amount of computation and ultimately determine the optimal path under a finer environment model through the A* algorithm. Regarding traffic management [19], Xu et al [20] proposed a framework based on AIS data analysis, including a historical traffic analysis module, K-means based attribute classification, and short-term traffic prediction module based on Back-Propagation Artificial Neural Network, to improve the efficiency of operations management in the Vessel Traffic Service.…”

Revolutionizing Marine Traffic Management: A Comprehensive Review of Machine Learning Applications in Complex Maritime Systems

“…Security Threats: The maritime industry faces threats such as piracy, smuggling, and illegal fishing [24][25][26]. Addressing these threats and ensuring the security of vessels and their cargo is a considerable challenge.…”