The Future of Sonar Processing

For three decades, the marine survey industry has accepted a painful truth: collecting data is fast, but processing it is slow. Autonomous Underwater Vehicles (AUVs) can run missions around the clock, while processing teams sort through the data at a fraction of real-time, manually correcting bottom tracks, tweaking gain curves, splitting projects because software can't handle full datasets, and slogging through hundreds of contacts that turn out to be nothing. Now, with the SonarWiz Fathom, that era is over.

The Math is Simple

Calculate what is spent today on sonar processing between labor hours, expensive workstations, shipping hard drives across the country because files are too large to transfer, delayed deliverables, and lost bids because of lengthy turnarounds.

Now calculate what happens when processing time drops—when one processor handles the workload that used to require three, deliverables ship in days instead of weeks, and teams spend time on interpretation instead of babysitting a computer.

The Old Way is in the Past

There’s no use in pretending otherwise—many sidescan workflows were designed when vessels towed fish at two knots and processors had days to review each survey. Old data architecture assumes that human labor is the solution to every problem.

Meanwhile, current AUV fleets can collect more data in one week than a team processes in one quarter. Clients want deliverables in days, not months. Processors are burning out, buried under backlogs that grow faster than they shrink.

The tools haven't kept pace with the sensors and the workflows haven't kept pace with the platforms. The industry built rockets but kept the horse-drawn carts. With Fathom, this can change.

Designed for AI From the Ground Up

Fathom isn't legacy software with automation bolted on as an afterthought. It's architected from the foundation as a platform ready for machine learning, designed from day one around the premise that machines should handle what machines do best, freeing humans for what humans do best.

We're developing deep learning models trained on real-world sidescan and synthetic aperture sonar data spanning diverse seafloor conditions. Rocky terrain, gaseous sediments, steep slopes, altitude excursions—the failure modes that send traditional rule-based algorithms off the rails become training data for models to learn what the bottom actually looks like, not just what rules define it.

Fathom's processing engine analyzes acoustic response patterns across the entire survey and applies adaptive corrections that balance imagery automatically, indicating intelligent gain normalization, no more manually adjusting TVG curves and consistent mosaics without the tedium.

Fathom supports automatic mosaicking at scale—cloud-parallel processing stitches thousands of lines into georeferenced imagery. Teams can upload the data, walk away and return to a finished product.

Built for the AUV Era

Surface-towed sidescan surveys aren't disappearing, but they're no longer the frontier. The frontier is autonomous—vehicles running multi-sensor payloads at depths and distances that surface vessels cannot reach.

These platforms generate data at rates that legacy workflows cannot absorb. A single AUV mission produces what used to take a survey vessel a week. Fleet operations multiply that by factors of five, ten, or even twenty. The backlog isn't a temporary inconvenience. It's a structural failure of the traditional processing model.

Fathom was designed for this reality, capable of:

• Direct ingest from any platform. Drag and drop from the vehicle's storage media, with no format conversion gymnastics or intermediary software. Raw data in, processed data out.

• Parallel cloud processing. Why process one line at a time when Fathom can process hundreds simultaneously? Fathom distributes workload across scalable cloud infrastructure. Processing time becomes a function of compute allocation, not human availability.

• Real-time QC during operations. Surface intervals become processing windows. Data streams to Fathom, preliminary products generate automatically, and operators assess quality while the vehicle is still in the water. Reacquisition decisions happen in the field, not weeks later in the office.

• Massive datasets under one project, whether its sidescan, SAS, or magnetometer. Terabytes of multi-sensor data from week-long AUV campaigns live in a single unified project. No splitting surveys across multiple files because software can't handle the volume. No stitching together outputs from different applications.

From Detection to Decision

Finding contacts is easy. Any threshold algorithm can generate a list of anomalies. The hard part—the part that consumes human hours and determines operational outcomes—is deciding which contacts matter. Legacy workflows dump hundreds of detections on processors and call it automation. Fathom challenges this.

Machine learning contact classification: We're developing neural network models trained on labeled imagery to distinguish mine-like objects from geology, debris from targets, threats from clutter. The goal is to cut false alarm review by an order of magnitude while maintaining detection performance.

Confidence scoring: Detections carry probability estimates, high-confidence targets rise to the top, marginal detections get appropriate scrutiny and processors allocate attention where it creates value.

Multi-sensor correlation: Fathom integrates sidescan contacts with magnetometer anomalies in a unified environment. No more switching applications to cross-reference sensors or manual coordinate matching. The data converges because the platform was designed as such.

Cloud-Native, Collaboration-Ready

The future of processing isn't a desktop application running on one workstation in one office. It's a platform that meets data where it is and delivers results where they're needed.

Teams can process anywhere—if you have a browser, you have Fathom. Collaboration happens in real time, meaning a processor in one time zone and an analyst in another work the same project simultaneously. Processing power and cloud infrastructure can scale to demand, without limits. Security meets individual requirements, from government-grade encryption and access controls to audit logging.

Don't Get Stuck in the Past

The organizations that thrive in the next decade of marine survey will be the ones that recognize a shift in what's possible. AUV fleets are scaling, data volumes are exploding and clients—navies, energy companies, infrastructure operators—expect faster answers at lower cost.

Things can continue the way they’ve been—manually correcting bottom tracks, tweaking gain curves by hand, and watching competitors deliver faster. Or, you can look to the future of sonar processing with SonarWiz Fathom.