Imagine trying to train a pilot or a computer system for a rare emergency, but you don't have any real recordings of that event happening. This project creates a 'flight simulator' for data, using AI to invent realistic scenarios that look and behave like real air traffic. It allows systems to learn from thousands of fake but accurate examples without needing to wait for real-world accidents or delays to occur.
AI tools for air traffic management are limited by a lack of available high-quality data, especially regarding rare safety-related events. This makes it difficult to train models that generalize well across different airports and scenarios.
A Universal Time Series Generator (UTG) and a Universal Time Series Forecaster (UTF) that create high-fidelity synthetic ATM datasets using GANs and VAEs.
If you are an airport operator dealing with unpredictable flight delays and congestion — this project developed AI models that generate synthetic datasets. This allows you to simulate new airport environments and predict traffic flow without needing years of historical data.
If you are a software developer dealing with a lack of safety-related data to train your AI — this project developed a Universal Time Series Generator. This tool creates high-fidelity synthetic datasets to test your automation tools against rare but critical safety events.
If you are a regulator dealing with the need to validate new airspace management rules — this project developed validated synthetic datasets and evaluation metrics. This enables you to stress-test system resilience and safety standards in a virtual environment.
Based on available project data, no pricing or commercial cost structure is mentioned; the project was funded by an EU contribution of EUR 993,550.
Yes, the project developed a Universal Time Series Generator (UTG) designed to generate synthetic data for a new airport simply by using a compressed representation of that environment.
Based on available project data, specific licensing terms are not provided, though the project involved a consortium of 4 partners including SINTEF AS.
The project provides recommendations for integrating synthetic data into existing ATM frameworks to improve predictive analytics and decision-making.
The project period runs from 2023-09-01 to 2026-02-28, indicating the tools are currently in the development and validation phase.
The consortium is a lean group of 4 partners across 4 countries (BE, IT, NL, NO). It features a balanced mix of research and industry, with a 25% industry ratio (1 industry partner, 1 SME, 1 university, and 2 research organizations), led by SINTEF AS. This structure suggests a strong focus on translating academic AI research into practical aviation tools.
Contact SINTEF AS in Norway
Contact us to explore licensing the Universal Time Series Generator for your aviation data needs.
