How Do Uncrewed Ocean Drones Work?

On 19 September 2026, a seven-metre vessel with a rigid wing sail left Hobart, Tasmania, and turned south. Nobody was aboard. Over the next five years it and three sister craft will circle Antarctica repeatedly, each circuit covering around 12,000 miles, measuring the exchange of carbon dioxide between the Southern Ocean and the atmosphere through winters when no research ship goes there.

That mission is a good illustration of what these vehicles are actually for. They are not faster, more capable or more comfortable than a research ship. They are slower, carry far less and cannot do most of what a ship does. What they offer is the ability to stay somewhere for months at a time, cheaply, in conditions that would be unpleasant, expensive or dangerous for people, and that single trade explains every design decision visible in the diagram.

 

Why It Is a Sailing Vessel

 

The wing sail, the solar panels and the underwater keel are all answers to one question: how do you keep a vessel at sea for six months without refuelling it?

A rigid wing sail works on the same principle as an aircraft wing rather than a cloth sail, generating lift from airflow and converting it to forward thrust. It needs no crew to trim it, holds its shape in any wind strength, and consumes no fuel. The keel beneath provides the lateral resistance that lets the vessel sail at an angle to the wind rather than simply being blown downwind, and its ballast rights the craft if a wave knocks it over.

Solar panels do not propel the vessel. They power the sensors, the navigation computer and the satellite link, which is a much smaller energy demand than propulsion and one that a few square metres of panel can meet. The division is deliberate: wind for motion, sun for electronics, and no diesel for either.

The result is extraordinary endurance. Saildrone Explorers have completed missions exceeding six months and covering more than 20,000 kilometres. The cost is speed. A Saildrone Surveyor crossed from San Francisco to Hawaii, around 3,650 kilometres, in 28 days, which works out at an average under three knots. A research ship does that passage in five or six days. For a vessel whose purpose is to loiter, this is an acceptable trade. For anything that needs to arrive somewhere quickly, it is disqualifying.

 

The Power Budget Problem

 

Here is the constraint that governs what these vehicles can actually do, and it is the most useful thing to understand about them.

Endurance and energy are different quantities. Wind and solar give a vessel unlimited duration but a very small power budget, and ocean instruments vary enormously in how much power they draw. Passive sensors measuring temperature, salinity, pressure or humidity consume almost nothing. Active acoustics, which create sound in order to listen for its return, consume a great deal.

The practical consequence is a split in the fleet. The wind-and-solar Explorer carries oceanographic and atmospheric sensors and a single-beam echo sounder, but it does not have the power budget to run a multibeam sonar. For seafloor mapping, one of the four applications in the infographic, a larger vehicle is required: the 20-metre Surveyor carries a 30 kilohertz EM-304 multibeam, a higher-frequency EM2040, EK-80 fisheries sonars and an eDNA sampling system, and it does so with an auxiliary diesel-electric system whose engine must run periodically to keep the batteries charged. Other survey USVs such as the 7.7-metre DriX are straightforwardly diesel-powered, trading endurance for the ability to run sonar continuously at up to ten knots.

So the vessel in the infographic, with its wing sail and solar panels, is a climate and weather platform. The seafloor mapping application on the same page belongs to a different, larger, partially fuelled member of the family. Both are uncrewed drones; they are not interchangeable.

 

What It Measures, and How the Data Gets Home

 

The sensor list in the infographic is accurate and the vehicles deployed on NOAA's hurricane missions give a concrete version of it: wind speed and direction, air temperature, sea surface and subsurface temperature, relative humidity, barometric pressure, salinity, and wave height and period. Two of the 2026 hurricane fleet additionally carried NOAA ASVCO2 sensors measuring carbon dioxide exchange across the sea surface.

Getting that home is where a real limitation sits, and it is rarely mentioned. Satellite links are not equal. Iridium services provide reliable global coverage with enough bandwidth for situational awareness, command and control, and a trickle of summary data, which is exactly what a climate mission needs because temperature and pressure readings are small. Full-rate transmission of survey data is a different proposition: marine broadband radio handles it well but only out to around 20 kilometres from a receiving station, which is why survey USVs often work within line of sight of a mother ship and send their data at the end of each survey line. Low-earth-orbit constellations may change this, and were still being evaluated at the time of the most detailed published assessments.

This is why a multibeam survey drone and a weather drone have such different operating patterns. A multibeam sonar generates data far faster than any satellite link can carry it, so the vehicle must either stay near a receiver or store the data and deliver it physically. A sensor reading sea surface temperature every minute produces a volume that fits comfortably down a narrow pipe from anywhere on Earth.

 

Into the Hurricane

 

The capability that best demonstrates the value proposition is the one no crewed vessel would attempt.

Saildrone has deployed uncrewed vehicles into Atlantic hurricanes alongside NOAA every year since 2021, scaling from five vehicles in 2021 to seven in 2022, twelve in both 2023 and 2024, and ten in 2026. Across those seasons 21 different vehicles have intercepted 21 named hurricanes and tropical storms on 46 separate occasions, accumulating more than 2,600 days on hurricane research missions. The hurricane-configured vehicles use a shorter, ruggedised wing designed for the wind and sea states inside a tropical cyclone.

The scientific target is rapid intensification, defined as an increase in maximum sustained winds of at least 30 knots within 24 hours, and it remains among the hardest things to forecast. Harvey in 2017, Ian in 2022 and Idalia in 2023 all intensified rapidly shortly before landfall, which is the worst possible timing for evacuation decisions. The physics driving it happens at the sea surface, in the exchange of heat, moisture and momentum between a warm ocean and the storm above it, and that boundary is precisely where aircraft cannot sample and ships cannot safely be.

The stakes justify the effort. NOAA's National Centers for Environmental Information attribute 52.9 percent of US billion-dollar disasters between 1980 and 2024 to tropical cyclones, with inflation-adjusted damages exceeding 1.5 trillion dollars.

 

Five Years Around Antarctica

 

The Southern Ocean mission launched in September 2026 addresses a gap of a different kind. The project, Constraining Ocean Carbon with Optimized Observing, is led by NOAA, Columbia University and Australia's CSIRO with the University of Washington and South Africa's CSIR, funded by Schmidt Sciences. Four Explorers will be deployed in sequence, with AI-driven ocean and atmosphere models selecting courses that route them through high-value sampling areas, and each completed circuit followed by a return to port for repair and redeployment.

The ocean absorbs roughly a third of the carbon dioxide humans emit each year, and the Southern Ocean is estimated to account for around 40 percent of that uptake. It is therefore one of the most consequential numbers in the global carbon budget, and it is measured poorly, because ships rarely go there in winter. As NOAA's Adrienne Sutton put it, the project fills a critical knowledge gap by adding winter observations.

What makes this more than a gap-filling exercise is what an earlier circumnavigation suggested. A 2019 Saildrone voyage indicated that in winter, carbon-rich water brought to the surface may release carbon dioxide rather than absorb it, which if confirmed at scale would revise estimates of how much carbon the Southern Ocean actually takes up. That is a finding only continuous winter presence could produce, and it is exactly the category of question these vehicles exist to answer.

 

The Cost Case, Honestly Stated

 

The economic argument is strong and worth stating with its caveats attached.

A large offshore research vessel costs over 100 million dollars to build and frequently exceeds 50,000 dollars a day to operate, with crew costs often accounting for more than half of that figure. A diesel survey USV costs a few million dollars, with operating costs estimated around 10,000 dollars a day including the shore-based operators, though published assessments note these day rates are still being established. NOAA has estimated at least a 40 percent gain in survey efficiency from deploying a single USV.

The caveats matter. No current system is genuinely autonomous; every one of these vehicles is piloted remotely by people, and the shore-side operations centre is a real and ongoing cost rather than an afterthought. Vehicle size is constrained by what a mother ship can launch and recover, which limits sensor size and therefore survey depth. Range is limited by communications as much as by endurance. And one of the most experienced researchers in the field has cautioned explicitly that the costs, capabilities and limitations of these systems remain poorly understood, and that overpromising risks a backlash when they fail to deliver what was claimed.

The sober version is that USVs do not replace research vessels. They do the long, dull, repetitive and dangerous parts of ocean observation at a fraction of the cost, which frees ships for the work that needs people, heavy equipment and deck space.

 

Who Pays

 

One dimension absent from most explainers is worth naming, because it is shaping the sector.

The same vehicle that measures sea surface temperature can detect, classify and track other vessels, and that capability has a customer with deeper pockets than ocean science. Recent contracts include a 10 million dollar US Navy award for Atlantic seafloor mapping, a 15.5 million dollar US Coast Guard contract, a renewed Navy counter-narcotics mission, a year-long Swedish maritime domain awareness deployment in the Baltic, and a strategic partnership with Palantir on AI-driven maritime intelligence.

This cuts both ways and both deserve stating. Defence and security revenue funds fleet expansion, manufacturing scale and reliability improvements that civilian science could never have paid for, and the Antarctic and hurricane missions benefit from hardware matured on those contracts. At the same time it means the platform's development priorities are increasingly set by surveillance requirements rather than oceanographic ones, and that questions about data access and mission tasking become more complicated. The Baltic deployment in particular sits in the same waters where undersea cable damage has repeatedly raised attribution questions, which is a reminder that an ocean sensor and an ocean surveillance asset are physically the same object.

 

More of the Ocean, for Longer

 

The infographic's closing line is precisely right, and the strength of it lies in the word longer. These vehicles are not better than ships at anything except persistence, and persistence happens to be what ocean observation has always lacked most.

The ocean is under-measured not because instruments are poor but because putting people on it is expensive, so almost all historical observation has been snapshots: a ship transects a region for three weeks, produces excellent data, and leaves. The Southern Ocean in winter and the inside of a hurricane are the two clearest cases of places where snapshots have never been available at all. A vehicle that costs a fraction as much and can simply stay changes what questions can be asked, and that, rather than any particular sensor, is what the technology actually delivers.

 

Note: This article reflects the state of the sector as of November 2026, drawing on sources including NOAA's announcements of the COCO2 Southern Ocean mission and the 2026 hurricane season deployment, Mayer's technical assessment of uncrewed surface systems published by the International Hydrographic Review, and reported contract awards. Cost figures for uncrewed vessel operations are estimates and remain under active revision. Companion pieces in this series compare AUVs, ROVs and USVs, and cover ocean data buoys.