The boat your AIS can’t see: Why we need to really understand the new navigation tools

Stuart Greenfield

Oct 15

19 min read

This article is updated from an earlier FirstShift publication when we had just 200 subscribers with nearly 1000 today I have re-written and added to it.

In brief

Dark, but not stormy! New electronic tools are transforming safety at sea, if you use them well

• AIS is a cooperative system. It shows the boats that choose to transmit, that are required to, and that you can hear. Fishing boats under 15 metres, most leisure craft and warships may not be on it at all, and the IMO says so in writing.

• It is not real time. A Class B yacht reports every 30 seconds. At a 20 knot closing speed that is over 300 metres between updates, and CPA is a projection, not a prediction.

• Having every tool is not the same as seeing. In 2018 a Volvo 65 with radar, AIS and a lookout, crossing a well lit fishing boat, still hit it. A fisherman died.

• The picture is about to get much wider. Solid-state Doppler radar now draws around 17 W, machine vision cameras sailed on 25 of the 40 IMOCAs in the last Vendée Globe, and VDES, the successor to AIS, enters SOLAS on 1 January 2028.

• More tools do not mean a true picture. Every tool misses something. AIS misses boats that don’t transmit, radar misses a person in the water, and a camera misses anything more than a few hundred metres away. Because they miss different things, you carry more than one, and somebody still has to look out of the boat.

• The rule that matters dates from 1972. COLREGS Rule 7(c): “Assumptions shall not be made on the basis of scanty information, especially scanty radar information.”

The night we were beaten by a light

In the autumn of 1983 I came home from Newport, Rhode Island, where the Australians had just taken the America’s Cup with their winged keel. I had been crew on the British yacht Victory with Peter de Savary, the team boss. That winter I made the final of the RYA’s Crew Match scheme, one of the few real doors into serious racing at the time, did not win it, and went back to my day job in the Royal Navy.

Which is how, the following summer, I came to get a call from Pam Saffery Cooper asking whether I would join her and her husband Brian aboard Dragon.

Dragon was a proper boat: an Ed Dubois one tonner, part of the British team that won the 1981 Admiral’s Cup alongside Robin Aisher’s Yeoman XXIII and Peter de Savary’s Victory of Burnham. No passengers. I was brought aboard to helm and contribute tactically, which in practice meant saying very little and trying hard not to be the reason we lost.

I had the chance to race on Dragon - Top boat in its day

The race was the RORC Cowes to Dinard, as everybody still called it, because before the marina was built the fleet anchored off the Dinard yacht club after the finish and waited for the tide to open the lock into the basins at St Malo.

We had no GPS, no plotter, no GRIB files, no phones and no satellites. We had paper charts, a B&G log, an echo sounder, a pencil, and whatever we could see.

As dusk fell we did what everybody did then: picked a boat ahead and locked on. A rival was the only performance instrument that told you anything about the conditions you were actually sailing in.

It was a beat, cold and wet, with the Casquets still 45 miles away. I had the helm and a four hour watch in front of me. Ahead, a scatter of stern lights; to either side, reds and greens we were passing quickly. The one that mattered was a white stern light I was fairly sure belonged to a fast French one tonner, because I had caught her sail number before the light went.

White. Steady. Just far enough ahead to matter.

Then, very slowly, it began to fade.

First you doubt yourself. Then you check the trim. Then the heading. Then you start making small, slightly desperate changes, because the conclusion is obvious and you do not like it: he is sailing away from you, and you are on the helm.

There were comments on deck, a great deal of trimming and many suggestions. I suspect they put it down to the new boy. We pushed harder, re-trimmed, re-thought everything, and nothing stopped that light getting dimmer and dimmer until it slipped over the horizon and was gone.

Brian came up and asked how we were doing. There was a long silence before somebody told him we had lost the boat ahead. He did not seem troubled. He let me keep the helm and quietly checked the speed and the trim, and said nothing.

Only later did we find out what had probably happened. The boat ahead had fitted a dimmer switch to her stern light, and somebody aboard was winding it down, just slowly enough, to look as if she were pulling away. I think Brian knew the trick already. We never proved who they were, or whether they beat us. It was, after all, a gentleman’s sport.

It did not change reality. It changed perception, and perception changed what we did.

That is the most useful lesson about electronic awareness I have ever been taught, and it was taught by a stern light.

The dimmer switch went digital

Navigation lights gave you three or four miles. Beyond that, boats ceased to exist, and your picture of a race was built from fragments: a light, a bearing, a sail number glimpsed at dusk.

Then came Decca, then GPS, then AIS, and suddenly we could see everything. Name, course, speed and closest point of approach at fifteen miles. It felt like certainty.

The early years offshore were nothing of the kind. Some boats had it, many did not, and some in the racing fleet ran it happily all day and then reached for a switch at the chart table at dusk. The boat you had tracked all afternoon simply vanished. Out of range? Failure? Tactics? You could not know. It was the dimmer switch again, rebuilt in software.

The sport has since closed that door, and closed it in almost exactly the right words. The Sailing Instructions for the 2025 Rolex Fastnet Race required every boat to have “their AIS transponder switched on, transmitting (at least every minute) and receiving” from 0900 on race day and throughout the race, and then added a line that might have been written with Dragon‘s rival in mind:

“Nothing shall be done to inhibit AIS transmissions.”

So the gamesmanship has gone. The problem has not, because the problem was never really the other boat. It was what we assumed the light, or the screen, was telling us.

What AIS actually is

Strip away the screen and AIS is a radio conversation: every transponder broadcasts its own position, course and speed on two VHF channels and listens to everybody else. Three facts follow, and almost every bad AIS decision I see comes from forgetting one of them.

It is line of sight. Range depends on the height of both aerials and the power of the transmitter. A ship’s 12.5 W Class A set, high on the superstructure, is heard a long way off. A race boat’s Class B set at a couple of watts, sharing a masthead aerial through a splitter, much less far. You are not looking at the racecourse. You are looking at a moving bubble around your own boat, and it is a different size for every target in it.

It is not real time. Each set reports at an interval set by its class and its speed. The IMO’s own guidelines, and the manufacturers’ manuals, carry the same table:

Look at the bottom line, because it is probably your boat and most of the boats around you. Two yachts closing at a combined 20 knots cover a little over 300 metres between reports. Worse, the IMO notes plainly that “Class B devices operate at a reduced reporting rate or when free time slots are available.” In a crowded channel, the yacht is the one that waits its turn.

So the closest point of approach on your plotter assumes both vessels hold their course and speed until it happens. Offshore, at night, with a fishing boat working its gear or a racing yacht about to tack, that assumption is often wrong. CPA is a projection, not a prediction.

It is over the ground. AIS reports course and speed over the ground. Your racing decisions are made through the water. In the Channel Islands or the Solent, with three or four knots of tide under everybody, the vectors are honest about the seabed and silent about the race.

Who isn’t on the screen

AIS is cooperative. If a vessel is not transmitting, or you are not receiving it, it does not exist on your display. A quiet screen does not mean empty water. It means incomplete information.

The IMO is unusually blunt about this. Its guidelines for the operational use of AIS on ships, Resolution A.1106(29), warn the officer of the watch that “other ships, in particular leisure craft, fishing boats and warships, … might not be fitted with AIS.” The same document says AIS “does not replace, but supports, navigational systems such as radar target-tracking,” and that “the user should not rely on AIS as the sole information system.”

That is the shipping industry’s own regulator, writing for professional watchkeepers on the bridge of a ship. Now think about the view from a yacht’s cockpit at 0300.

Small fishing boats. In European waters, the carriage requirement for fishing vessels starts at 15 metres overall. The EU directive phased it in by size between 2012 and 2014, and it stopped there. The crabber working pots off Start Point, the trawler inside the Casquets, the open boat off the Raz: many of them are under fifteen metres and under no obligation to transmit anything at all.

The dark fleet. In January 2024 a team led by Fernando Paolo of Global Fishing Watch published satellite radar and imagery analysis in Nature showing that more than three quarters of the world’s industrial fishing activity is missing from public tracking systems, and nearly a third of transport and energy vessel activity. Most of the gap is in Asian and African waters rather than ours, and that is the honest counterweight. But the principle travels. As Paolo put it: “We can’t manage what we can’t see.”

The ships that are where they are not. AIS position comes from the transmitting vessel’s own GNSS receiver, which means anything that fools the receiver fools everybody listening to it. In May 2025, ship tracking services showed a large circle of Russian vessels in the Gulf of Finland that the Finnish Coast Guard said simply was not there. I wrote about GPS jamming and spoofing in July. AIS inherits every one of those vulnerabilities, and adds the possibility that somebody is lying about who they are as well as where.

Weak signals and wet aerials. Even boats that do transmit are not always heard. The independent report into night racing in congested waters that followed the 2018 Hong Kong collision found that “several boats experienced unreliable and degraded performance with the fitted AIS system”, and recommended a better masthead connector and aerial and a proper testing regime. Those were professionally prepared Volvo 65s.

Every tool, and still a collision

If you want to know why this matters, read what happened to Vestas 11th Hour Racing just after 0100 on 20 January 2018, about thirty miles from the finish of Leg 4 of the Volvo Ocean Race, on the approach to Hong Kong.

This was not a careless crew or an under-equipped boat. Mark Towill, skippering that leg, later gave his own account to Yachting World. “I was at the navigation station monitoring the radar and AIS,” he said. “I was watching three vessels on AIS. There were a number of additional boats on AIS, many of them fishing vessels.” On deck, “the on-deck crew confirmed visual contact.” The night was dark and cloudy, about 20 knots of breeze, a moderate sea. The target was not hard to see: “the fishing vessel was well lit.”

“The crew confirmed we were crossing the fishing vessel when, before the anticipated cross, there was an unexpected collision.”

A fisherman died. I do not tell that story to second guess anybody aboard. I tell it because it is the clearest example I know of a boat with every layer working, radar, AIS and human eyes, and the picture still being wrong in the one place it mattered. The independent report found that the fishing fleet off Hong Kong “did not form an impenetrable barrier”, and that virtually all of the vessels carried some lighting and AIS. The tools were there. The information was there. The assumption about what the other vessel would do next was the thing that failed.

That is what COLREGS Rule 7(c) has been saying since 1972, in fifteen words that every racing navigator should have on the chart table:

“Assumptions shall not be made on the basis of scanty information, especially scanty radar information.”

Replace the word radar with AIS and nothing about the sentence needs to change.

The same report made a second finding that matters for the rest of this article. The Volvo 65s carried an FMCW radar, which gave good discrimination between close targets but weak range: “Radar detection ranges were about a third of that expected from a conventional 4 kW pulse radar.” It recommended replacing it, and it recommended looking hard at a masthead thermal camera, because “a small camera producing a stabilised picture could soon be fitted at the masthead.”

That was 2018. Soon has arrived.

Over the horizon: what is coming aboard

There are four new layers arriving on racing boats, plus one old one that is quietly coming back. None of them is a silver bullet, and I will try to be as honest about their blind spots as about their brochures. Brands here are examples, not endorsements.

1. Radar comes back, smaller and hungry for less

Barking Mad, my Swan 40, carried a Furuno 1721, a standalone CRT set from the 1700 series. The screen lived down below, and from the cockpit you could just see it through the hatch over the aft cabin. Between 1999 and 2006, before AIS, I only switched it on when we were cruising.

Barking Mad, 1999, full on Cowes Week, radar and pole is in the container!

The night it earned its keep was a passage into Alderney in fog, with a south-westerly force 6 blowing and a spring tide running. It got us in. It was not magic. The screen was full of clutter, and it needed time and patience to set up properly before it showed you anything worth trusting, which is not a luxury you have when the tide is doing the navigating for you.

Racing, it stayed off, and for a generation most of the racing boats I sailed on did without. Weight aloft, windage, a power-hungry magnetron and a warm-up delay were poor trades on a boat built to be light, and AIS seemed to make them unnecessary. That is a judgement from watching fleets, not a census.

The technology has changed underneath that habit. Solid-state, pulse compression radars have replaced the magnetron in the small radome market. Raymarine’s published specification for its 18 inch Quantum 2 Doppler dome is a fair example of where the class now sits: a 20 W peak output, 17 W at maximum transmit, 7 W on standby, 5.6 kg, instant on, ranges from 6 metres out to 24 miles, and tracking of up to 25 MARPA targets at once. Doppler processing colours closing targets red and opening targets green, which is precisely the question you are asking at 0300: is that one coming towards me? It also goes some way to answering the problem I had with the Furuno off Alderney, which was never seeing returns. It was deciding which of them mattered.

Seventeen watts at full transmit is well under the 25 to 40 W that Practical Boat Owner measured for the Starlink Mini. The power argument against radar has largely gone. Weight and windage have not, so I would expect it back first on bigger boats and in the two-handed fleet.

There is also a rule-shaped reason. COLREGS Rule 7(b) says: “Proper use shall be made of radar equipment if fitted and operational.” If you carry it, you are expected to use it, including “radar plotting or equivalent systematic observation of detected objects.” Fitting a dome you switch off to save the batteries is the worst of both worlds.

Radar’s blind spot is the mirror image of AIS’s. It sees the unlit fishing boat that is not transmitting. It does not see a person in the water, a semi-submerged container or a whale with any reliability, and in rain and a big sea it fills with clutter. It tells you something is there. It does not tell you what.

2. Machine eyes at the masthead

The camera the Volvo report imagined now exists, and it went round the world on most of the IMOCA fleet. SEA.AI’s masthead units combine daylight and thermal cameras with software that compares what it sees against a library of objects, and raises an alarm. The company’s figure is that 25 of the 40 IMOCAs that started the 2024 to 2025 Vendée Globe carried one. That is a vendor’s number, not an audit.

The most telling account I have found comes from Jean-Marie Clément of Mer Concept, describing small craft encountered on the Transat en Double: “These boats were not visible on the radar and did not have AIS. The only way to spot them was through the camera, which alerted us to the danger of a collision.” Read that twice. Not on radar. Not on AIS. On the camera. It was published by SEA.AI’s own press office, so weigh it accordingly, but the physics it describes is sound.

The honest counterweights are price, range and night. When Yacht magazine looked at the cruising version in September 2023, the entry Offshore One unit was €10,990, with quoted detection of yachts at about 1,000 metres, small boats from 250 metres and people, buoys and smaller flotsam from about 100 metres. The racing unit’s own dealer specification gives a detection angle of 110 degrees in daylight and twilight, and 50 degrees at night. That narrowing after dark is exactly the time you most want the width.

So the camera is not long range awareness. It is the last line of defence, the electronic version of the bowman shouting. It is very good at the one thing radar and AIS are worst at: small, cold, unlit things close ahead.

3. AIS grows up: VDES and the satellite picture

AIS was built for ships to tell each other who they were, not for the traffic now riding on it. Its successor is VDES, the VHF Data Exchange System: the same two AIS channels, plus additional data channels and, crucially, a satellite component. The international spectrum for the satellite part, VDE-SAT, was allocated at the World Radiocommunication Conference in 2019.

This year it moved from engineering to law. In May 2026 the IMO’s Maritime Safety Committee, at its 111th session, adopted amendments to SOLAS Chapter V allowing ships to carry VDES in place of AIS. They enter into force on 1 January 2028. Carriage stays voluntary and ships come first. Small craft will follow when the equipment is cheap enough.

Two things will change for us when it arrives. First, a two-way channel to shore: navigational warnings, weather and safety information delivered over the same equipment. Second, a picture that is no longer bounded by the height of your aerial.

You can already buy a version of that second promise. Digital Yacht enabled satellite AIS messaging on its Class B+ transponders in July 2025, with a service showing other AIS targets within a hundred mile radius. I went through the physics of satellite AIS in some detail in “Always On”, so the short version here: it arrives in bursts, with far more delay than VHF, and it degrades worst in dense traffic, which is exactly where a racing fleet is thickest. It is a fleet picture, not a collision avoidance tool. Whether using it while racing counts as outside help is a separate argument, and that piece makes it.

4. Your crew on the screen

AIS has quietly taken on a second job that has nothing to do with ships: finding your own people.

World Sailing has required personal AIS crew overboard beacons in the higher offshore categories since the 2018 to 2019 Offshore Special Regulations. The current 2026 to 2027 edition goes further at boat level, with EPIRBs required to carry internal GNSS and AIS transmission, so the beacon that alerts the rescue centre also shows on every plotter within range. And from the start of 2025, European rules have pushed manufacturers towards beacons that combine AIS with a DSC alert to your own boat’s VHF.

Most personal AIS beacons live inside an automatic lifejacket and fire themselves when it inflates. Ocean Signal’s own description of its MOB1 is that “the action of the lifejacket inflating triggers deployment of the antenna and automatic activation of the beacon”, with the first alert inside 15 seconds. That is exactly what you want when somebody goes over the side.

It is rather less welcome when the bowman takes a big green one on the foredeck, a water-activated inflator decides that counts as immersion, and every plotter within range lights up with a crew overboard alarm for somebody standing on deck, soaked and cross. Somebody aboard needs to know how to switch that beacon off, fast. A hydrostatic inflator, such as the Hammar type, waits for real immersion rather than spray and cuts the odds considerably. On a boat that is going to be wet for days, it is worth the money.

It fires when the jacket inflates. Make sure somebody knows how to stop it.

The second half of the problem is the radio. In the 2021 Rolex Sydney Hobart, on the night of 27 December in Bass Strait, a personal locator beacon aboard Celestial was set off by accident. Nobody was in the water. The Australian Maritime Safety Authority alerted the race organisers. Ichi Ban, nearby, was asked to raise her, got no answer, and fired flares. The race committee protested that Celestial had been unresponsive on the radio for 90 minutes.

The international jury found that “Either the equipment or procedures on board Celestial resulted in her failing to maintain a continuous listening watch on VHF 16 as required by SI 31.4,” and added a 40 minute penalty to her elapsed time in lieu of disqualification. That moved her behind Ichi Ban, who took the Tattersall Cup, and who was given three minutes of redress for her trouble. The beacon was an accident. The silence afterwards cost them the race.

It is not a one-off. The Ocean Racing Club of Victoria counted more than a dozen inadvertent beacon activations across the Sydney Hobart and its own Melbourne to Hobart Westcoaster in a single December, and set up a formal investigation to find out why.

So the blind spot here is mostly behavioural. A beacon nobody has armed, tested, or coded to the boat’s MMSI is a beacon in a pocket. A beacon that fires by accident with nobody on the radio is a race committee problem. Test them at the start of the season with the boat’s plotter on, make sure everybody knows what the alarm looks and sounds like and how to cancel it, and keep a proper listening watch on 16.

5. The pipe, and what it does to the watch

Starlink does not see anything. It is not a sensor. But it changes the awareness picture in two ways, and both matter.

It makes the picture wider: live satellite AIS services, shore-based tracking and high resolution weather radar overlays all become available offshore, where a few years ago they needed a harbour Wi-Fi password.

And it changes the racing. Starlink changes the rules more than anything since GPS. If you can afford the data, and your batteries can carry it, weather and routing are always on, rather than something you pull down once a watch over a satellite phone and then live with. RORC already permits publicly available cloud-based routing services in its offshore races, so the question is no longer whether you may. It is whether you can pay for it and power it.

Both caveats are real. Starlink’s Personal Maritime plan arrived this summer at $185 a month, and ocean data on it, and on Roam Unlimited from 27 September, now costs $6 a gigabyte, three times the old rate. Practical Boat Owner measured the Starlink Mini at 25 to 40 W, continuously, which is more than the radar above at full transmit. Check the battery budget before you check the forecast.

Starlink: Always on weather and routing, for as long as the batteries and the budget last.

It also changes where people look.

My worry is not that Starlink is wrong. It is that it is compelling. A crew member scrolling a beautiful live fleet picture on a phone in the companionway is not keeping a lookout, however much information they are holding.

The fused picture problem

Put all of this together and a race boat can carry five overlapping layers of awareness: eyes, AIS, radar, a camera and a satellite picture. The next step is fusion: one screen, one target list, the radar return, the AIS identity and the camera image stitched into a single confident icon.

That is a real improvement, and I would like it on my boat. It is also the most dangerous version yet of the stern light, because the more convincing the picture, the less anybody questions it.

Here is the point that matters, and it is the reason to carry more than one layer at all. Each layer has a blind spot, and they are different blind spots.

AIS cannot see the unlit fishing boat that radar can. Radar cannot see the person in the water that a thermal camera can. The camera cannot see the ship at twelve miles that AIS can. And none of them can see what the skipper of the other vessel is about to do, which is the only thing that actually decided the night off Hong Kong.

That empty column is the one that matters. It is why COLREGS Rule 5 still begins, as it always has: “Every vessel shall at all times maintain a proper look-out by sight and hearing.” Sight and hearing first, and then everything else available.

How I would use it on a race boat

None of this is an argument for switching anything off. The tools are extraordinary and the boats that use them well are safer and faster. It is an argument for giving each one the authority it has earned and no more.

1. Treat AIS as early warning, not full awareness. It is the best long range tool you have for ships. It is a poor tool for small craft and a useless one for anything that does not transmit.

2. Assume something exists outside the screen. Especially near fishing grounds, headlands and harbour approaches at night. A quiet display is not empty water.

3. Never let a vector override the plan on its own. CPA on a 30 second update with tide running is a projection. Before you give away leverage or break your routing, get a second source: a bearing, a radar return, a look.

4. Delay decisions, but not beyond control. Most AIS conflicts resolve themselves if you watch two or three updates. Decide in good time how close is too close, and act decisively when you reach it.

5. If you fit radar, use it properly. Rule 7(b) expects systematic observation, not an occasional glance. Set guard zones, track the targets that matter, and know how your Doppler colours read before the night you need them.

6. Test the whole chain before the season. Transponder, splitter, aerial, beacons and the boat’s MMSI. The Volvo boats found degraded AIS performance. Assume yours has it until you prove otherwise.

7. Protect the lookout from the screens, and brief what they cannot show. One person outside, eyes up, at night in traffic. Tell the watch where the small fishing boats will be, and that they may not appear on the plotter. The screens are for the navigator. The sea is for the watch.

Sail the racecourse first. Manage the picture second.

The light was real… final thought

That stern light in 1984 was entirely real. It was exactly where it appeared to be, and every crew member on Dragon could see it. Nothing about it was false except what we decided it meant.

The screens are the same. The data on them is mostly accurate. The ship at twelve miles really is at twelve miles. The danger has never been that the picture lies. It is that it is incomplete in ways it never tells you about, and so convincing that you stop looking for the missing pieces.

The next generation of kit will make the picture wider, sharper and much more persuasive. Radar will come back, cameras will see in the dark, VDES will carry the picture over the horizon. All of it will help. None of it will tell you what the other boat’s helmsman is about to do, or that someone ahead has a hand on a dimmer switch.

Brian Saffery Cooper knew that forty years ago, and said nothing, and let me work it out.

Stuart Greenfield FirstShift — Yacht Racing Intelligence From first shift to first place.

ffshore racing is never just about the boat, the forecast, or the routing software.

It is about decisions made under pressure, in changing conditions, by tired people trying to keep the boat moving fast and safely in the right direction.

That is where races are won and lost.

At FirstShift, my aim is simple: to help owners, skippers and crews race offshore with more clarity, more confidence, and better preparation whether that is a RORC Channel race, a Fastnet campaign, or your first serious step into offshore racing.

Thanks for reading Firstshift - yacht racing intelligence! This post is public so feel free to share it.