A Survey on AIS-Based Vessel Trajectory Completion: Methods, Limitations, and Open Challenges

The Automatic Identification System (AIS) is an essential source for the analysis of maritime mobility, but its trajectories often contain noise, discontinuities, and temporal gaps. This review analyzes recent open-access approaches for repairing, imputing, and reconstructing AIS trajectories affected by interruptions or data losses. The review distinguishes between local methods, suitable for isolated errors or short losses, and contextual methods, based on historical AIS data, deep learning, generative models, or multisensor fusion, mainly oriented toward prolonged gaps. The analysis shows that the most effective solutions for long gaps usually depend on additional information, such as representative historical trajectories or traffic patterns, large volumes of training data, or external sensors. Therefore, the problem of reconstructing routes in areas with low AIS coverage, limited historical data, and no complementary observations remains open. In these scenarios, the challenge is not only to complete a sequence of points, but to generate a route that is navigable, consistent with vessel dynamics, and reasonable from an operational point of view.