System architecture

How HYDRA works

Hybrid Defect Recognition and Analysis

A unified sensor pipeline that produces a metrically accurate, fully navigated 3D digital twin in real time. Live on the surface, as the survey runs.

The HYDRA pipeline

Six subsystems. One coherent output. Hover over a node to explore each layer of the architecture.

Hover over a node to explore the HYDRA pipeline.

SLAM architecture

Not just SLAM. Proprietary multi-modal fusion

Most subsea SLAM systems rely on a single sensor modality. Stereo vision in clear water, or sonar when visibility drops. That means data gaps, manual stitching, and post-processing to fill in what was missed. HYDRA takes a fundamentally different approach.

HYDRA fuses stereo vision, multibeam sonar, DVL/INS, and depth simultaneously, in real time. When one modality degrades, the others compensate automatically. Periods of poor visibility don't break the model or lose position. The system maintains tracking through sonar and inertial data until visual conditions recover.

Positional drift, the single biggest challenge in subsea SLAM, is addressed through a proprietary correction methodology that maintains metric accuracy throughout the survey. Loop closure serves as validation, not correction. No post-processing alignment. The model is dimensionally accurate as it's built.

01

Stereo Vision

DeepSea SmartSight MV1000

Photogrammetric 3D reconstruction with colour texture. Primary geometry source in clear water conditions.

02

Multibeam Sonar

Oculus M750d dual-frequency

Geometric localisation in zero visibility. Structure-relative positioning and visual blackout continuity.

03

DVL / INS

Nortek Nucleus 1000

Velocity-aided navigation and attitude reference. Sensor fusion anchor for continuous positioning.

04

Depth

Pressure-based absolute Z reference

Compensates DVL altitude limitations and provides robust depth registration throughout the survey.

Recognition

Defect Recognition.

Utilising AI deep learning and object recognition, HYDRA automatically identifies and classifies defects directly from the fused sensor data as the survey runs. Corrosion, marine growth, anode depletion, coating loss, and structural anomalies are detected, measured, and georeferenced without manual intervention.

The models are trained on real subsea inspection data, not generic image sets. Every finding is flagged with position, measurement, and classification confidence, giving the reviewing engineer a structured dataset, not a video to scrub through.

This is not a post-processing step. Recognition runs live, in parallel with the SLAM pipeline. Findings appear in the model as the ROV moves through the structure.

Live

Real-time detection

AI

Deep learning classification

Temporal depth

AI defect recognition imagery
Deliverables

Instantaneous Deliverable

Every HYDRA survey produces a structured inspection report generated directly from the SLAM model and sensor data, not from handwritten field notes or retrospective video review.

Reports include automated CP potential assessment, marine growth classification and thickness estimation, corrosion mapping, scour depth measurement, dimensional survey outputs, and structural condition summary. Each finding is tied to a precise georeferenced location in the 3D model.

Every report is delivered in a structured, engineer-ready format, ready to review and act on. Consistent quality is a system property, not a personnel dependency.

CP Assessment

Potential mapping · Anode condition

Marine growth

Classification · Thickness · Coverage

Corrosion

Surface condition · Pitting · Coating loss

Scour

Depth · Extent · Trend over time

Measurements

Dimensional · Gap · Clearance

Structural

Condition summary · Anomaly register

Data management

Your single point of record

Every inspection dataset is stored in HYDRA's native cloud database, giving you a permanent, searchable record of every survey, every finding, and every measurement across your entire asset portfolio.

HYDRA can serve as the definitive point of record for subsea integrity data, or integrate directly with your existing integrity management system through API. Your data, your way, with no manual re-entry.

NATIVE

Cloud database

Full hosted solution. HYDRA becomes your integrity management database. All survey data, temporal records, and reporting accessible from a single platform.

API

Direct integration

Connect HYDRA outputs directly to your existing integrity management databases. Structured data, consistent format, no middleware required.

Ready to run HYDRA on your next campaign?

Talk to us about your asset, your survey window, and what you need the data to do.