Voyis: Enabling Subsea Autonomy with Complete Optical Payloads
Surveyors have historically been able to easily integrate into their vehicle platforms all the sensors needed to achieve their goals, simply connecting everything though the vehicle tether for surface control. Now however, the industry is seeking fully autonomous remote operations, a goal that requires sensors, navigation, and vehicle control systems to work together as a complex system to deliver unsupervised decision making. This can only be achieved when the sensors instantaneously provide the information and analysis necessary to enable true autonomy.It is an exciting time in the ocean sector where the combination of emerging robotics technologies and a global pandemic is driving a shift to uncrewed operations. The advent of autonomous systems, and a focus on remote operation, data automation, and smaller low-cost platforms is enabling more efficient and low-cost subsea operations that will open the door to bolder exploration of the depths. But these innovative concepts bring complex challenges that must be overcome.History of Optical Sensor DevelopmentVoyis led the way in advanced optical sensor development with the integration of its Insight Pro laser scanner into the C&C Technologies (now Oceaneering) HUGIN AUVs in 2012, bringing long range high-resolution 3D laser data and stills images to subsea pipeline inspections. With this new real-time quantifiable data, it enabled reliable pipeline tracking and automated defect detection that was not feasible with multibeam sonar. Because of these developments, optical sensors have now become standard practice for AUV pipeline surveys.The next step was incorporating these new optical sensors into Work Class ROVs for wide area 3D modelling with the goal of creating digital twins of subsea installations for monitoring assets and conducting dynamic subsea metrologies. In 2016, Voyis worked with Sonardyne to tightly integrate the Insight Pro laser scanner with the Sprint-Nav navigational solution to achieve incredibly accurate 3D models across vast areas. The ROV simply traverses the field, with no asset interaction or requirements to deploy static scanning equipment, and builds up a model of what it sees on the control computer. These instantaneous 3D datasets drastically improved survey efficiency and inspection confidence.Shifting to Real-time Data EnhancementEven with these incredible capabilities, a vision of fully autonomous operations requires reliable unsupervised decision making, by which the confidence in these decisions is completely dependent on the quality of the input data. Though laser scanning and stills imaging has always delivered real-time data when compared to photogrammetry, the output can vary with environmental conditions, and machine learning is most robust with consistent datasets. It is therefore imperative that the sensors themselves add real-time processing capability to enable dependable automation.Voyis is delivering on these requirements by both removing noise from point cloud data with advanced filtering algorithms, and instantly enhancing and correcting images as they are collected. Our robust machine-learning-based colour correction algorithms help us to see consistent true-colour photos of subsea targets. When combined with navigational data, it is possible to deliver real-time geo-corrected laser models, and consistent images across incredibly long surveys. This uniformity produces wide area consistent image mosaics without variations in quality from image to image. Adding onboard analysis software, like EIVA’s Onboard Deep Learning, enables the automated detection of pipeline defects and features, to drastically reduce analysis time.A color image mosaic of subsea pipeline. Image courtesy VoyisMiniaturization into Complete SolutionsThe final and most significant challenge in subsea operations has remained constant, the cost and inaccessibility of ocean exploration. For our future to see a proliferation of optical sensors documenting our oceans, these high-resolution sensors must be delivered in packages suitable for small, low-cost vehicle platforms that lessen the required deployment infrastructure. The level of complexity now demanded requires not just piecemeal sensors, but complete payload solutions with synergistic integrations between sensors, navigation solutions, and vehicle platforms.To this end, Voyis has launched the Recon line of AUV payloads, developed in partnership with vehicle manufacturers for common platforms like the HII REMUS 100 and L3Harris IVER 4. The hydrodynamic modules mount in-line with the vehicle using standard mechanical interfaces, connect to existing electrical inputs, and are controlled with the vehicle’s mission planning interface. All data can be saved to the payload’s onboard hard drive to simplify data management and the vehicle’s navigational data is accessed automatically to geotag images and spatially correct 3D laser data to improve target localization.Optical payloads are now being combined with side-scan sonar from Wavefront, onboard deep learning from EIVA, and a contactless cathodic protection system from Ocean Floor Geophysics to deliver application specific payloads. This reduces barriers to adoption and makes it easier than ever before for surveyors to adopt new survey technology.Recent innovations in highly capable small vehicle platforms, miniaturized sensor technologies, and advanced sensor autonomy provide the building blocks to achieve the future vision of fully remote operations. But true autonomy is not achieved in isolation, it will require surveyors, vehicle designers, and sensor manufacturers all working together in partnership to overcome the challenges, and together, Illuminate the Unknown. Image courtesy Voyis