The work this is built on.

ApertureLab is not a first attempt at this problem; it is the current state of a line of work that runs from acoustic wave scattering through to the models that read the resulting imagery. The record below is 29 peer-reviewed publications across 14 years, and it covers the whole chain rather than one stage of it.

A dark poster showing 64 stacked sonar ping traces rising into a seabed feature, titled the seafloor, knowable
Sixty-four consecutive returns from one seabed feature. A single ping says almost nothing; coherently combined, they resolve a place.

Research that makes this possible.

ApertureLab's physics engine is grounded in peer-reviewed sonar science: GPU-accelerated TDBP beamforming, precision interpolation kernels for sub-wavelength micronavigation, and validated sub-bottom acoustic simulation.

AI & Machine Learning

Physics-accurate synthetic data is the missing ingredient for underwater AI; ApertureLab generates it. The research arc below runs from synthesizing scarce training data with a physics-coupled GAN, to few-shot classifiers trained on synthetic augmentation, to zero-shot VLMs that classify sonar imagery with only a text prompt.

Reed, Gerg et al. 2019
MTS/IEEE OCEANS 2019

Coupling Rendering and Generative Adversarial Networks for Artificial SAS Image Generation

Albert Reed, Isaac D. Gerg, John D. McKay, Daniel C. Brown, David P. Williams, Suren Jayasuriya

A physics renderer coupled with a GAN produces realistic SAS target imagery when labeled field data is too scarce to train from alone.

Read paper →
Gerg, Lynch & Blanford 2025
MTS/IEEE OCEANS 2025

Low-Shot Learning for Synthetic Aperture Sonar Image Classification Using Hierarchical Pretraining & AirSAS

Isaac D. Gerg, Alex Lynch, Thomas E. Blanford

Hierarchical pretraining on AirSAS synthetic data reduces labeled-data requirements to a handful of real examples per class.

Gerg 2026
IEEE IGARSS 2026

Prompted, Not Trained: On Zero-Shot Classification of Synthetic Aperture Imagery with Vision-Language Models

Isaac D. Gerg

VLMs classify SAS targets at 0.946 AUC using only text prompts describing highlight-shadow geometry; zero domain-specific training required.

To appear.

The Full Research Arc

The cards above are highlights. The complete record is 29 publications across 14 years, and it spans the entire chain: physics-based simulation and synthetic data, micronavigation, GPU image formation, learned autofocus, compression, perceptual image quality, and the models that read the imagery. The diagram below places each paper on the stage of the processing chain it advances.

29 publications, 2012 to 2026, mapped onto the SAS processing chain; each line ties a paper to the stage it advances.

Next: the physics engines, or why a synthetic aperture exists at all.