Skip to main content
Category

Data

Large-Area Optical Surveys: The Future of Subsea Inspection

By Autonomy, Data

The Need for Change in Subsea Surveys

The demand for large-scale seabed surveys is growing rapidly, but traditional survey methods face critical challenges: 

  • Data overload: Advancements in sensor technology mean we now collect vast amounts of data from sonar (multibeam, side-scan, synthetic aperture sonar) and optical systems (laser, images, video), making efficient processing essential. 
  • Slow and costly processing: Data processing remains largely manual, requiring significant time and expertise. 
  • Shortage of skilled personnel: The industry lacks enough qualified professionals to process the increasing volumes of data. 
  • Inefficiency in workflows: Existing workflows were designed for linear asset inspections (e.g., pipelines) but must now scale for wide-area surveys. 

The challenge is not just acquiring more data but extracting meaningful insights faster and more efficiently. Addressing these inefficiencies is key to transforming subsea surveying.

The Shift to Autonomous Underwater Vehicles (AUVs)

AUVs have revolutionized underwater data collection by enabling high-speed large-area surveys. However, the shift to AUVs has primarily increased the rate of data acquisition rather than solving the core problem of data processing. 

Despite their advantages, AUVs must still return to the surface before data processing can begin. The reliance on post-mission analysis means that while we are collecting more data than ever, we are not necessarily gaining insights any faster. The bottleneck remains: a lack of automated, efficient processing to convert raw data into usable information in real time.

The Cathx Approach: Accelerating Insight Generation

Cathx has developed a structured approach to solving this challenge by focusing on three key stages.

Stage 1: Reducing Data Volume to What Matters

The first step is prioritization—reducing the workload for skilled operators by minimizing the volume of data they need to review. To achieve this, we are developing Clarity microservices that use advanced machine vision techniques—such as colour and shape detection—to automatically filter data and highlight events of interest. For example, an environmental scientist monitoring coral health can receive only the images where coral structures are detected and flagged for signs of discoloration or bleaching, allowing them to focus on meaningful changes rather than combing through irrelevant footage.

These filtered datasets not only streamline analysis but also create high-quality inputs for training AI models. As more labelled data is collected, model accuracy continues to improve, enabling increasingly precise and automated detection of critical subsea events.

What Is CATHX Clarity?

Clarity is the world’s first autonomous perception platform for subsea data, designed to automate data processing workflows directly on underwater vehicles—at the moment data is collected. At its core is a modular microservices architecture: each Clarity microservice performs a specific processing task, such as sharpening images or detecting objects, and these can be combined into custom workflows tailored to any mission. With Clarity, organizations can build scalable, real-time solutions that transform raw data into actionable insight—fast, efficient, and ready for AI.

Stage 2: Bringing Processing to the AUV

To further enhance efficiency, event detection should move onboard the vehicle itself. By embedding real-time processing capabilities within the AUV, we can identify critical data as it is collected. Once the AUV is retrieved, decision-makers can immediately review key insights rather than waiting for extensive post-mission processing. 

This shift significantly reduces the turnaround time for large-area surveys. Important discoveries—such as structural anomalies, environmental changes, or asset integrity issues—can be flagged as soon as the AUV surfaces, allowing for immediate decision-making.

Stage 3: AI-Driven Automation

The final step is leveraging machine learning to enhance event detection and classification. Initially, AI models can be trained offline using high-quality labelled datasets. Over time, these models can be deployed directly on AUVs, enabling onboard classification of features and anomalies without human intervention. As more data is collected, it becomes possible to further improve the performance of AI models. 

Cathx’s roadmap is designed to facilitate AI model development and integration, allowing users to build their own detection capabilities and integrate them seamlessly as Clarity workflows. Beyond data classification, the ultimate goal is to enhance autonomous operations by integrating perception-driven 

mission planning. By enabling AUVs to adapt dynamically based on real-time data analysis, we move closer to true underwater autonomy.

Future Perspectives: Where Is Large-Area Surveying Headed?

The future of subsea surveying lies in automation, real-time processing, and AI-enhanced decision making. As sensor technology continues to evolve, the industry must shift away from traditional, linear workflows designed for localized asset inspections and embrace scalable solutions that can handle vast amounts of data and increasingly complex environments. 

Key advancements we anticipate include: 

  • Real-time adaptive mission planning: AUVs that can adjust their survey paths dynamically based on detected events 
  • Enhanced onboard processing: More powerful embedded computing capability to reduce reliance on post-mission data handling 
  • Greater interoperability: Seamless integration of optical and sonar data for comprehensive multi-modal analysis 

By embracing these innovations, we can transform large-area optical surveys from a time-consuming process into streamlined, intelligent workflows that deliver actionable insights faster and more efficiently than ever before.

Making subsea data smarter, faster, and more effective

The industry is at a pivotal moment—advancements in data acquisition must be matched by equally sophisticated processing and automation solutions. At Cathx, we are committed to redefining subsea survey workflows by reducing data overload, enabling real-time event detection, and driving AI powered automation. The future of large-area optical surveys is not just about collecting more data, but about making that data work smarter, faster, and more effectively for industry applications.

Stay in touch with our latest news and updates.

    The Data Bottleneck: A Growing Challenge for Autonomous and Uncrewed Operations

    By Autonomy, Data

    As the demand for subsea data grows, I’ve seen companies investing in autonomy and uncrewed operations struggle to scale how data is processed. 

    While autonomous underwater vehicles (AUVs) and other advanced systems are transforming data collection, there’s an untapped opportunity to automate the workflows that happen once data is collected. Heading into 2025, we now have the capability to scale the data processing of survey operations and complete the transformation to highly efficient subsea data workflows that can keep pace with today’s increasing complexity, volume, and frequency of data. 

    This fundamental shift from manual, reactive workflows to automated, scalable systems presents an enormous opportunity to modernize data processing workflows to the advantage of organizations on the vanguard of adoption.

    The Cost of Unscalable Data Processing

    Regardless of the industry or the project focus, a common refrain I hear is that scalability of data processing is a major inhibitor of progress, growth, and even project success. In defense, for instance, there’s a pressure to conduct larger and more frequent surveys to monitor critical infrastructure such as subsea cables and pipelines, and these assets are under heightened scrutiny due to their strategic importance. At the same time, evolving regulations and international mandates drive the need for more comprehensive environmental impact assessments covering increasingly vast areas. But in every case, delivery of actionable insights depends on the ability to quickly process the increasingly high volume of data gathered. 

    For every day of data collected, it can take three to five days to process that data.

    With data processed manually—subject to a finite talent pool and growing competition for expertise—the status quo has been an unavoidable data bottleneck that limits operational efficiency and delays insights. Scaling operations to meet the demands of emerging markets like environmental monitoring and infrastructure surveys has been nearly impossible with this model, but that old paradigm is now changing with the ability to automate the processing of the data.

    The cost of manual data processing is more than inefficiency—it’s a barrier to growth, timely insights, and scalability. In fast-evolving industries, traditional workflows create bottlenecks that strain resources, delay decisions, and limit our ability to meet market demands. Embracing automation is no longer optional; it’s essential to eliminating these constraints, enabling faster, more accurate insights, and positioning organizations to lead in a data-driven world.

    The Problem Is Three-Fold

    The challenges with subsea data collection and processing are multifaceted, creating significant barriers to scalability and efficiency.

    1. Manual Data Processing

    Most subsea data workflows rely heavily on manual techniques. For example, sonar and optical data still need experts to manually tag and classify objects of interest from dense datasets. For every day of data collected, it can take three to five days to process that data. This lag in delivering results is compounded by the risk of human error and subjective interpretation, which can lead to inconsistencies in data quality and reliability.

    2. Lack of Suitably Qualified Resources

    The subsea industry faces a critical shortage of qualified data processors and analysts. Interpreting complex datasets like multibeam sonar and optical imagery requires specialized skills—but the talent pool is small and in high demand. This shortage creates delays, increases costs, and limits growth.

    3. Lack of Scalable Workflows

    Current methods and tools for data processing were designed for smaller-scale tasks, such as pipeline inspections or localized seabed mapping. However, emerging applications—like large area environmental surveys or surveillance of critical infrastructure at higher frequency—require workflows that can scale to analyze higher volumes of data. Without scalable solutions, companies are unable to meet growing demands, stalling operational expansion and innovation.

    Reengineering Our Approach

    To bring efficiency to how we tackle the data processing challenge, we need to start by understanding the workflows for each individual process. By mapping out each input and the methodology employed to achieve the required outputs, we can then rebuild the process as an automated and optimized pipeline that minimizes manual intervention and maximizes data accuracy, speed, and value. 

    At Cathx, we are reengineering data processing workflows with automation at their core. Once automated, we can then execute these in parallel to current workflows to ensure seamless integration and validation. This parallel execution allows us to compare results, fine-tune processes, and address any discrepancies before fully transitioning to the automated system. By taking an iterative approach, we can build confidence in the accuracy and reliability of the new workflows while maintaining uninterrupted operations. 

    And this is the key to unlocking how to derisk and accelerate the development of autonomous systems. By creating robust, validated workflows that integrate seamlessly with advanced technologies, we ensure that autonomous systems operate with precision and reliability in complex underwater environments. This approach not only mitigates risks associated with adopting new technologies but also shortens the development cycle, enabling faster deployment of innovative solutions. At Cathx, our focus on automation is paving the way for next-generation autonomy.

    The Valley of Machine Learning Disillusionment

    We’ve all seen the hype around machine learning’s potential to transform subsea operations. But there’s a reality check: many companies are stuck in what I call the “valley of machine learning disillusionment.” They expect ML to deliver rapid automation and smarter decision-making, but the path to this end is complex. Unlike generic AI models, subsea operations require bespoke workflows, high-quality labeled data, and specialized models tailored to multi-sensor inputs like sonar, optical, and electromagnetic data.

    One of the biggest reasons machine learning hasn’t yet delivered in our industry is the lack of sufficient, high-quality data from the seabed. Without good training data in sufficient quantities, machine learning hasn’t had a chance to work.

    To move beyond this valley, we need a rethink. Standardizing methodologies, adopting shared automation frameworks, and focusing on incremental improvements can deliver better outcomes. It all starts with collecting larger amounts of higher-quality data in the first place.

    What’s the True Prize of Autonomy?

    Addressing the data processing bottleneck will unlock the full potential of autonomous surveying in subsea industries, creating substantial opportunities for service providers, AUV manufacturers, and sensor manufacturers to accelerate growth and innovation.

    For service providers, it’s now possible to move beyond traditional, project-based revenue models tied to the number of survey days, which are often unpredictable and resource intensive. Instead, service providers can adopt a service-oriented approach, offering long-term, subscription-based contracts where clients receive continuous access to high-quality, real-time data insights. 

    This model will provide more stable recurring revenue streams, support scalable growth, and enhance operational predictability, allowing service providers to plan resources more effectively while reducing costs. Almost more importantly, it will position companies as long-term strategic partners rather than one-off service providers, fostering deeper client relationships.

    For AUV manufacturers, addressing this challenge will significantly enhance the value proposition of their products. Real-time edge processing, combined with advanced perception capabilities like event detection and automated feature extraction, makes AUVs smarter and more efficient. This differentiation boosts demand, drives sales, and expands market opportunities.

    It also allows manufacturers to integrate new in-mission capabilities, enabling AUVs to autonomously respond to environmental changes, adjust mission parameters, and optimize survey outcomes. These advantages will position AUV manufacturers as leaders in the rapidly evolving subsea technology landscape.

    The subsea industry is evolving rapidly, creating unprecedented opportunities and challenges. At Cathx, our mission is to help organizations overcome the data bottleneck, reduce operational risks, and develop sustainable, scalable solutions. By rethinking traditional workflows and introducing tools like real-time edge processing and modular microservices, we empower companies to process data faster, reduce costs, and future-proof their operations.

    We’re dedicated to enabling the full potential of autonomy and machine learning in subsea operations—helping companies make smarter decisions, improve efficiency, and maintain a competitive edge in an increasingly data-driven world.

    Stay in touch with our latest news and updates.