ESVAGT Gives the Officer on Watch a Few Minutes of the Future

ESVAGT is fitting NextOcean's WavePredictor to the service operations vessel ESVAGT NJORD at the Dudgeon offshore wind farm this month — a motion radar and software package that predicts what the vessel will do over the coming minutes — and it is the clearest example in this issue of an AI system aimed at improving a human decision rather than replacing or grading it. What it does: the system combines motion radar with software to deliver real-time vessel motion predictions for the coming minutes, supporting personnel transfer and operational go/no-go decisions in challenging sea conditions. NextOcean CEO Karel Roozen put the operational reality plainly: "On an SOV, the crew makes dozens of go/no-go calls a day. Our system shows the officer on watch what the vessel is going to do in the coming minutes, so that call is backed by verified predictions and not by feel alone." At Dudgeon the commercial driver is direct — technicians need frequent turbine access whenever weather permits, so every avoidable no-go is a lost working day and every misjudged go is a person stepping between two moving objects. Why the design is the lesson. Compare it with the camera systems in our first story. A behaviour-monitoring camera produces a record of what a person did, reviewed later, by someone else, often against a target. WavePredictor produces information the person needs at the moment they need it, and the decision stays theirs. Both are AI applied to safety. Only one of them increases the operator's authority rather than the office's oversight. When crews resist safety technology — and they do — this distinction explains most of it, and it is a design choice, not an inevitability. Why this matters for ship managers beyond offshore wind: the same class of prediction applies wherever a motion window governs a decision — pilot transfers, crane operations, launch and recovery, heavy-lift, STS operations, gangway use in a swell. The offshore wind sector is funding the maturity because it has hundreds of transfers a day and a hard commercial number attached to each one. What to do now: list the decisions on your vessels where someone judges a motion window by feel and where being wrong is an injury. That list is your candidate set for this technology, and it is probably shorter and more valuable than the list your vendor would propose. Then ask any supplier of prediction tooling the only question that matters: what is the prediction horizon, and what is the measured error at that horizon? The honest caveat: no accuracy figures, prediction-horizon specification, workable-day increase, transfer-count data or contract value has been published. A single vessel at a single wind farm is a deployment, not evidence, and ESVAGT has published no before-and-after comparison. Wave and motion prediction over a few minutes is a well-founded physical problem rather than a speculative one, which is a point in its favour — but "verified predictions" is the vendor's phrase, and no independent verification of them has been described. [Smart Maritime Network]

This story is part of the Maritime AI Digest of 30 August 2026.