The problem
Vessels do not move randomly. Over time, historical AIS trajectories reveal recurring patterns of movement — established shipping lanes, commonly followed routes, port-to-port corridors, and characteristic movement patterns associated with different vessel types. Cargo ships, tankers, and passenger vessels tend to follow well-defined routes, while fishing vessels often exhibit distinctive spatial clustering and activity patterns in their operating areas.
This raises an important question: can we automatically learn these normal movement patterns from historical vessel trajectories and identify, in real time, when a vessel deviates from them?
Our approach
We have developed a real-time route extraction and anomaly detection system that learns normal maritime movement patterns from historical trajectories and continuously evaluates vessels against these learned patterns.
Our algorithms extract generalized vessel routes and behavioral clusters based on factors such as vessel type, historical movement, geographic region, and route frequency. These learned patterns form a dynamic representation of normal maritime behavior. As new AIS data becomes available, the system continuously compares a vessel's current trajectory against the expected route or behavioral pattern.
Detecting deviations in real time
The system can identify:
- Route deviations — vessels departing significantly from commonly followed shipping routes.
- Behavioral anomalies — movement patterns that differ from the vessel's expected behavior or that of similar vessels.
- Fishing activity clusters — recurring spatial patterns and operating zones associated with fishing vessels.
- Emerging route patterns — changes in the way vessels move that may indicate a broader shift in maritime traffic.
- Changes in established routes — detecting when previously stable traffic corridors begin to change over time.
Rather than relying on predefined routes alone, the system learns routes from historical data and continuously adapts as maritime traffic evolves.
From historical data to real-time intelligence
The result is a dynamic maritime monitoring system that can answer, in real time:
By combining trajectory mining, route extraction, vessel behavior modeling, and real-time anomaly detection, our solution transforms historical AIS data into an intelligent early-warning system for detecting unusual vessel movements and emerging changes in maritime traffic patterns.