Innovate UK backs Rovco-led project to create 3D models from live ROV video data
Readers may remember Bristol-based subsea technology firm Rovco from last year’s Subsea Expo issue, in which we profiled the company’s adoption of 3D photogrammetry, also known as image-based modelling (IBM), and its applications in the subsea industry.
2017 saw the company develop the process, which involves using high-resolution photos of a static object or site area taken from multiple angles and processed to generate a 3D vector model from point cloud data, with funding from technology agency Innovate UK.
Now Rovco’s latest project, also with backing from Innovate UK, will see it develop a 3D visualisation system for collected data, as part of a two-part artificial intelligence (AI) demonstrator project worth up to GBP1 million (US$1.4 million).
Working with the Offshore Renewable Energy (ORE) Catapult, the first phase of the project will involve building and validating the equipment and software necessary to produce live 3D data from subsea environments. This will be trialled and tested at ORE Catapult’s renewable energy test facility in Blyth.
The second phase will then include the development of a complete 3D vision-based survey solution, part-controlled by AI. According to the partners, this technology could reduce offshore inspection costs by up to 80%, owing to advances in camera technology and embedded graphic processing, and the ability to mount these systems on smaller and more intelligent autonomous underwater vehicles (AUVs).
The first phase of the project will be 70% supported by Innovate UK, with the remaining 30% funded by Rovco. The planned phase two is also expected to be supported by Innovate UK once technical feasibility is proven. This will see Rovco partner with an AUV manufacturer and other subsea technology providers.
3D from video
Rovco chief executive and founder Brian Allen said: “Compared to traditional visual survey methods, 3D delivers precise measurements and reliable metrics for asset condition monitoring. Combining this with the use of AI to better analyse and understand subsea data enhances asset integrity decisions and reduces the manpower required, while also speeding up project turnaround times.”
In particular Allen highlighted the potential cost savings the technology could realise in the subsea inspection and decommissioning sectors.
Speaking with InnovOil via email, Allen added: “Phase 1 will prototype and demonstrate the feasibility of a high-quality, underwater, calibrated, stereo camera system with embedded computing to enable innovative real-time processing of underwater 3D from ROV video survey.”
Allen said that recent advances in both camera technology and embedded GPU computing would enable the team to produce a system “impossible even a year ago,” and builds on its previous Innovate UK-funded study.
ORE Catapult’s wave and tidal sector specialist, Simon Cheeseman, said: “The Catapult is in a unique position to not only identify opportunities for the use of disruptive technologies in offshore wind farm operations and maintenance to improve efficiencies and reduce costs, but also to help and support innovative companies to develop, demonstrate and validate these technologies using our unique testing facilities.”
Phase 2 will then test this equipment in real conditions, Allen said, with a view to developing deep-learning based autonomous survey capability, using 3D data. “This brings another step-change in addition to the 3D capability, with autonomous condition assessment and difference modelling,” he added. “Machine assessments of condition provide repeatable, dependable, consistent measures rather than relying on varied human assessment.”
The project is emblematic of much of the current state-of-the-art work in the subsea and inspection sectors at the moment, where machine learning is proving itself to be a game-changing addition to existing tools. InnovOil looks forward to covering the results of Rovco and ORE’s efforts.