IRAV2: Industrial Cleanup of Contaminated Sites in Dumping Areas 2

At a glance

• Development of a highly automated system for the detection and classification of underwater unexploded ordnance (UXO) in dumping areas
• Combination of modern sensor technology (dual-frequency SAS, seismic, fiber-optic acoustics) with AI-supported analysis and synthetic training data
• Measurement campaign with harbor and at-sea tests for validation under real-world conditions
• As part of the IRAV2 collaborative project, Fraunhofer IWES is responsible for the INVERSA subproject, which aims to achieve data-driven, real-time seismic characterization and classification of silted-over unexploded ordnance. A multichannel seismic towed array system developed in the predecessor project IRAV is to be specifically expanded during the project period and prepared for real-time data acquisition. AI-accelerated waveform-based inversion methods will be used for data analysis, enabling physically motivated UXO object characterization and classification. Finally, the inversion results are intended to contribute to multisensory data fusion.

The challenge

Large quantities of corroding munitions lie in the North Sea and the Baltic Sea. Existing methods are highly labor-intensive, produce high false alarm rates, and reach their limits, particularly with objects buried in sediment. To date, there has been no highly automated workflow - from detection to classification - for the safe and cost-effective clearance of larger marine areas.

The solution

IRAV2 builds on the predecessor project IRAV and develops an integrated sensor system with automated analysis: A dual-frequency synthetic aperture sonar (SAS) mounted on an autonomous underwater vehicle (AUV), an advanced multichannel seismic system, and fiber-optic acoustics are being tested in a joint measurement campaign. AI methods and generative AI models generate and utilize synthetic sonar data to improve object classification. Fraunhofer IWES provides realistic seismic simulations, high-resolution waveform inversion methods, and a real-time analysis pipeline that directly derives material and location properties of unexploded ordnance (UXO) from the measured data.

The added value

The implementation of the planned multi-modal measurement and analysis pipeline will lead to faster and more reliable surveying of dumping sites. The combination of sensors and seismic-based data fusion reduces false alarm rates, improves risk assessment, and lays the foundation for large-scale, cost-effective clearance campaigns. It enhances maritime safety, reduces environmental and infrastructure risks, and can thus strengthen Germany’s position as a technology hub in the field of offshore safety and critical infrastructure.