Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Atlantes: A system of GPS transformers for global-scale real-time maritime intelligence

Henry Herzog, Joshua Hansen, Yawen Zhang, Patrick Beukema

TL;DR

The paper addresses the need for real-time, global-scale maritime intelligence derived from AIS GPS streams. It presents Atlantes, a transformer-based system with a continuous point embedding layer and CNN front-ends, trained on large, expert-annotated datasets to classify entity type (vessel vs buoy) and vessel activity (transiting, anchored, fishing, moored, other). Key results show high entity accuracy (≈97.5%) and competitive activity accuracy (≈71%, with 90% for fishing), enabling real-time insights across a daily throughput of billions of GPS messages. By open-sourcing the models and tooling, the work aims to democratize access to global maritime intelligence and support timely governance and interventions.

Abstract

Unsustainable exploitation of the oceans exacerbated by global warming is threatening coastal communities worldwide. Accurate and timely monitoring of maritime activity is an essential step to effective governance and to inform future policy. In support of this complex global-scale effort, we built Atlantes, a deep learning based system that provides the first-ever real-time view of vessel behavior at global scale. Atlantes leverages a series of bespoke transformers to distill a high volume, continuous stream of GPS messages emitted by hundreds of thousands of vessels into easily quantifiable behaviors. The combination of low latency and high performance enables operationally relevant decision-making and successful interventions on the high seas where illegal and exploitative activity is too common. Atlantes is already in use by hundreds of organizations worldwide. Here we provide an overview of the model and infrastructure that enables this system to function efficiently and cost-effectively at global-scale and in real-time.

Atlantes: A system of GPS transformers for global-scale real-time maritime intelligence

TL;DR

The paper addresses the need for real-time, global-scale maritime intelligence derived from AIS GPS streams. It presents Atlantes, a transformer-based system with a continuous point embedding layer and CNN front-ends, trained on large, expert-annotated datasets to classify entity type (vessel vs buoy) and vessel activity (transiting, anchored, fishing, moored, other). Key results show high entity accuracy (≈97.5%) and competitive activity accuracy (≈71%, with 90% for fishing), enabling real-time insights across a daily throughput of billions of GPS messages. By open-sourcing the models and tooling, the work aims to democratize access to global maritime intelligence and support timely governance and interventions.

Abstract

Unsustainable exploitation of the oceans exacerbated by global warming is threatening coastal communities worldwide. Accurate and timely monitoring of maritime activity is an essential step to effective governance and to inform future policy. In support of this complex global-scale effort, we built Atlantes, a deep learning based system that provides the first-ever real-time view of vessel behavior at global scale. Atlantes leverages a series of bespoke transformers to distill a high volume, continuous stream of GPS messages emitted by hundreds of thousands of vessels into easily quantifiable behaviors. The combination of low latency and high performance enables operationally relevant decision-making and successful interventions on the high seas where illegal and exploitative activity is too common. Atlantes is already in use by hundreds of organizations worldwide. Here we provide an overview of the model and infrastructure that enables this system to function efficiently and cost-effectively at global-scale and in real-time.
Paper Structure (11 sections, 10 figures, 1 table)

This paper contains 11 sections, 10 figures, 1 table.

Table of Contents

  1. Introduction
  2. Methods
  3. Tasks and Datasets
  4. Model architecture and training
  5. How good is this model really?
  6. Deployment
  7. Conclusion/Impact Statement
  8. Appendix
  9. Change point detector logic summary
  10. Annotation Tool
  11. Additional Examples of Activity Predictions

Figures (10)

  • Figure 1: (A) Depiction of the Automatic Identification System. (B) All broadcasted messages from 2023 ($\sim$ 15B) color coded by vessel category.
  • Figure 2: Examples of activity from each class (A-D). Example buoy (E) and vessel (F).
  • Figure 3: ATLAS model architecture (SOG: speed over ground, COG: course over ground).
  • Figure 4: Example classifications of vessel movement patterns (Indian ocean, February 2025).
  • Figure 5: A. Depiction of annotation process. GPS sequences are annotated at message level granularity by assigning a class to each individual message.
  • ...and 5 more figures