This presentation explores how vision-based sensing is evolving subsea inspection and positioning workflows, alongside the introduction of 3D Recon OC, a compact subsea 3D model reconstruction system designed for observation-class ROVs.
3D Recon OC is the newest addition to the 3D Recon product line, delivering the same high-resolution, geospatially accurate 3D model outputs in a smaller, more deployable architecture. This enables applications such as pipeline surveys, asset inspection, and integrity analysis to be carried out using observation-class ROVs, extending metrology-grade inspection capabilities to more efficient platforms.
Building on this foundation of spatial perception, the presentation will also highlight developments in vision-based positioning through systems trained using deep learning techniques for object detection and pose estimation. S-NDAPS enables real-time understanding of subsea structures without the need for fiducial markers, supporting precise vehicle positioning, guidance, and the progression towards autonomous intervention tasks.
Crucially, the development of these systems reflects a shared learning between inspection and autonomy. The same vision and inertial data used to generate accurate 3D models are now being leveraged to train perception systems that allow subsea vehicles to interpret and interact with their environment.
Together, these approaches demonstrate how vision and inertial sensing can move beyond traditional inspection, providing reliable environmental understanding and enabling new methods of subsea positioning and control. This progression from reconstruction to perception-driven autonomy represents a key step towards more advanced and efficient offshore operations.