Safinah Roundtable Discussion | Round 2
AI, Robotics and Coatings: Is the industry ready for what comes next?
In this round, Carl Barnes, Simon Daly, Michael Hindmarsh and Mark Welch come together to discuss what Artificial Intelligence (AI) and robotics could mean for the future of coatings.
Drawing on their industry experience, the panel considers where these technologies could add real value, where their limitations remain, and how they may change the relationship between digital tools and human technical expertise.
How far could AI and robotics ultimately transform coating inspection – and where should human expertise remain firmly in control?
Watch the full Safinah Roundtable Discussion video as the panel explore AI, robotics and the future of coating inspection across marine, yacht, offshore and infrastructure assets.
Round 2 Discussion Highlights
Protective coatings play a critical role in safeguarding ships, offshore assets and major infrastructure from corrosion and deterioration. Yet coating inspection still relies heavily on visual assessment, manual data review and the judgement of experienced inspectors – processes that can be time-consuming, difficult to standardise and challenging in hazardous or inaccessible environments. Artificial intelligence and robotics could begin to change that.
Computer vision systems are increasingly capable of analysing inspection photographs and video to identify coating failures, assess corrosion coverage and classify coating condition. Instead of relying solely on visual judgement, AI could help turn inspection imagery into more measurable and consistent condition data.
The opportunity becomes even greater when AI is combined with remote inspection technologies.
Drones, robotic crawlers, remotely operated vehicles and magnetic climbing robots can collect images and sensor information from locations that may be difficult or dangerous for inspectors to reach. AI can then analyse that information for signs of corrosion, coating degradation and structural anomalies – potentially improving safety, coverage and the speed at which inspection findings are produced.
Coating Inspection is only one part of the story
AI could also analyse historical reports, measurements, photographs and maintenance records to identify deterioration patterns and track changes between inspection cycles. Over time, this could support a move towards more predictive, condition-based maintenance rather than relying primarily on fixed inspection intervals.
So where does this leave the human coating experts?
The emerging model may not be one of replacement, but augmentation. Human expertise remains essential for validating anomalies, interpreting complex coating failures, assessing specifications and making engineering decisions. The document points towards a human-in-the-loop approach, where AI handles initial screening while experienced professionals remain responsible for critical judgement.
Challenges also remain around data quality, standardisation, regulatory acceptance and integration with existing inspection systems.
Where AI still falls short
AI can support coating inspection, but it cannot replicate everything an experienced inspector sees, measures and investigates on site.
Image-based AI works from two-dimensional information. A human inspector works in three dimensions and can directly examine defects, measure DFT, investigate blister contents, assess defect depth and, where necessary, cut into or remove coating to determine the root cause. Many coating failures have multiple interacting causes that even high-resolution imagery may not reveal.
Drones and robotic crawlers also have practical limitations. They can be valuable where access is difficult, but connectivity can be unreliable in tanks and confined spaces, while crawlers may be slow when covering large coated areas. Their effectiveness also depends on the quality and range of the sensors they carry.
Standards present another challenge. Current inspection standards are largely based on the assumption that data collection, assessment and onsite measurements are performed by people, meaning AI-enabled inspection still sits outside many established frameworks.
Sensors introduce a different risk. A reading may produce a false positive or false negative because it represents only a small, localised area of a much larger structure. If AI interprets that result without sufficient context, it could lead to an inappropriate and potentially costly maintenance decision.
For example, a sensor on an offshore wind structure may indicate elevated corrosion risk, while the issue could actually be caused by local coating thickness, mechanical damage or environmental exposure.
For Safinah, the position should remain clear: AI, drones, crawlers and sensors are valuable tools, but they should complement – not replace – experienced “boots on the ground” coating inspectors. Watch the full video
Subscribe to our newsletter and be the first to know when the new video goes live.
If you have further enquiries and opinion about the topic, speak with our experts – contact us at enquiries@safinah.com

