Inside NATO’s Silent Underwater Submarine Hunters | The Military Tech Show
Nov 15, 2026
Inside NATO’s maritime research centre in Italy, scientists reveal silent underwater gliders that can operate at sea for months. Using ocean sensors and sensitive underwater microphones, these autonomous robots gather vital data and listen for nearby ships and submarines, potentially paving the way for a new generation of underwater surveillance.
AI Generated·May contain errors
The oceans of the world are one of the last places
that military technology can be hidden
from prying eyes. Satellites may be able
to see down to a few square centimetres on land,
but even assets as big as nuclear submarines can
be hidden in deep water. And even on the surface,
the world's seas are so massive
that finding vessels set on course trouble
or performing illegal operations can be extremely difficult.
Well, NATO's Centre for Maritime Research and Experimentation
is looking for solutions to problems on the world's oceans,
and the scientists there are leading the way.
It's very silent,
the way they move.
They are almost undetectable.
I am a senior machine learning scientist.
A very effective platform for my countermeasures.
The centre is based at La Spezia Naval Base in northern Italy.
Among its leading scientists is John Potter.
So John, first of all, just tell me
a little bit about what you guys do.
It's the only maritime science
and technology centre that NATO actually owns.
And our job is to deliver the best maritime science
and technology to the navies of the nations.
And just give us a sense of practically what that means.
Technology is always some kind of race,
and so there are always new things coming over the horizon.
And what we tend to work on is relatively game-changing technologies
that in 10 or 20 years will be out there on the water
assuring our defence and security.
And why is it that the centre can do these things,
that the individual nations, 28 of them?
It can't do.
If you're looking at game-changing technologies,
they're normally fairly high risk.
Some of them don't pay off, some of them do.
The nations can spread the cost, spread the risk,
and all of them enjoy the benefit through
a collaborative research hub.
Plus, being 28 nations,
we get the benefit of having many diverse cultures and viewpoints
and ways of thinking about a problem.
Thank you very much for having me.
Among the autonomous systems being worked on at La Spezia are gliders,
underwater robots that tend
not to use their propellers.
One day they could be used as sub-aqua border guards.
Right now they deliver a wealth of information
on potential ocean battle spaces.
Well, I've moved inside to one of the labs at the centre where
a lot of the experimental work obviously takes place.
A lot of it surrounding these, the gliders.
Emmanuel, you can tell me a little bit more about them.
Exactly what are we looking at here?
OK, so the gliders, they
are actually autonomous underwater vehicles as any other vehicle,
but with a very specific feature.
Which is the fact that they don't actually use propellers by default.
They have a buoyancy engine.
So basically they change their weight to become light and heavy.
And using that they propel themselves going up and down in the ocean.
So it's very silent the way they move.
They are almost undetectable underwater.
And they can be equipped with multiple payloads
or essentially multiple sensors
that can that allow us to do a battle space characterization
in a way that is not risky for men.
And very efficient because they stay
in the water for long periods of time.
So we've got an example here
or several examples of the kind of payloads that this can carry.
Talk us through what we're looking at.
Yes, this is an example of a hardware piece
that goes inside of the hull of the glider.
And then this is the one that this stays outside.
It's essentially the external sensors that allow us to measure
the environmental properties like the temperature,
salinity or sound speed.
All of these are actually then processed inside of the platform.
We just transmit small pieces through satellites
so that most of the information is still kept inside of the platform.
This is called the smart sensing.
So they only deliver through the small radio links
or satellite links the vital information.
Another sensor that is actually now in development is this one.
This is what we call an ADCP,
so it's acoustic Doppler current profiler.
And so what it does, it sends out acoustic beams
and then through the back scattering
of the water column you can actually measure the currents,
so basically allows to measure
the actual water velocity along the water column.
And this is an example of a battery unit that
basically goes inside of these slots
and we can have multiple units of these battery packages on
board to extend the life of the platform
and allow us to operate with sensors that are more demanding.
How long can it operate for?
Well, they can operate two, three months quite easily.
They have similar platforms to these ones.
They've crossed the Atlantic
and even longer ranges in the Pacific.
But typically, for these sort of engagements that we do,
we operate within one to two months' time frame.
And on the glider behind you,
a different configuration at the front.
Yeah, sure.
And so, besides these sort of payloads
that fit inside of the hull of the platform itself,
We are developing and doing research to develop these more,
let's say, advanced systems
that are basically using hydrophones.
So the performance of these sensors is quite outstanding.
So we have a very high sensitivity to detect the minimum small noise
that exists in the water column.
Allowing us to identify the direction where the sound is coming from
and doing this we can actually get bearings
of potential anomalies in the
water and then identify submarines or ships that are actually nearby.
So Manuel, we've seen the hardware, we've seen the gliders.
Just talk us through a little bit more about how they work
and what kind of information they can collate.
These platforms, we exploit their richness
on being quiet and being persistent.
They are combined with ships and other assets
that might exist in the area.
Examples that we have of these results
are actually embedded, for instance, on this map.
This is already a...
A product where we inject the information coming out from the gliders.
The red spots indicate that it's deeper
and so basically this means that areas where the
submarines will be detected more easily.
As an example, we did recently a deployment of these
platforms here on the Ligurian Sea during
a period of two months where this information
was actually being collected.
This is an example of a map with the multiple tracks that these robots
operated so you can see the richness of the information
and the density of the information
being collected that is quite unique you really need
cannot actually do that with manned systems
this is an example where the the three
of the the platforms were actually equipped with the
hydrophones
And they were actually able
to detect sound sources that were anomalies,
that basically could be submarines or a ship,
that then will be triggering missions
to classify surface boats or ships
that were actually identified in an area and could be hostile.
And is the glider's simplicity the key to its success in a way?
Yes, sure, because it's the simplicity
that allows it to be actually quite modular,
so that you put multiple sensors on it,
and then the fact that it's very easy to operate.
Easy to deploy. Makes it quite unique.
How quickly can you use that information?
Yes, you have a full data download cycle in about half an hour.
So you can see that it's quite fast.
And so they can actually respond
to the requirements of operational planning.
For tactical usage, where you really need to have seconds,
that's still an area of research.
That they can actually detect signals still underwater
and to be able to surface and communicate, that takes some minutes.
And so tactical and organic implementation
of these platforms is an area
that we are exploring with different payloads,
underwater communications
and the different ways to provide alarm signals.
And what are the limitations of this system?
The big limitation is the way it propels,
like we discussed before with the buoyancy engine,
makes it to go slow.
So if there are strong currents,
it's not going to respond that easily.
And also to maneuver it from an area to another area,
so it takes some time because the speed will always be slow.
Eventually it might be overcome,
and we are now working on all these...
hybrid versions with combining the propellers
and the buoyancy engines but still is a
I think it's one of the major limitations on it
and we are actually looking in terms
of even increasing the not only the the battery package
but also the increase the size of the
platform so that it can actually bring more batteries looking
into different ways to
actually recharge batteries there
are already solutions where you have used
The basically the gradients in the temperature
in the ocean to charge batteries so the you cannot
you do that when the gliders go deep in the ocean in the deep ocean
so basically it's called they are called
the thermal gliders they can stay in the water for
For years.
So gliders could stay at sea for years
and their ability to deliver information is speeding up.
Right now they can give NATO nations knowledge.
In the future fleets of them might stand silent
guard off coastlines around the world.
How fast is the technology developing?
Quite fast. And I would say that every year we
are seeing new solutions on the engineering side.
And even here and ourselves in the lab, every year we come up
with new algorithms and solutions that are quite outstanding.
So it keeps everybody quite motivated.
Well that is it for part one
but coming up how big data can spot the ocean's bad guys
and why NATO's navies might just
be a little safer thanks
to a specialist team in Italy.
Welcome back to the Military Tech Show,
where we bring you the very latest on the innovations
that could change how the armed forces of the world operate.
In this show, in a world exclusive,
we're looking at the work of NATO's Centre for Maritime Research
and Experimentation, or CMRE.
Among the scientists there are teams taking
on the challenge of big data in order to help make
the world's shipping lanes safer places.
After all, no shipping, no shopping.
The world's oceans are big places
and there's all kinds of data that can make a difference
in making them safer. Until recently we've never had enough information
and information has been
precious and what little we've been able
to gather we've been able to sit and pour over
and try and extract the best from and now we're
in a very different kind of game where we are
drowning in information
and the problem now is to see the wood for the trees.
So there are some very interesting techniques being developed
in data mining, for example,
deep learning, working on artificial intelligence,
which can understand at some more fundamental level
what's going on in a picture of life,
which may be formed out of the fusion of many different
different kind of streams of diverse data.
So we're looking at algorithms
which are able to sift through all of that data
and recognize what is a normal picture of life.
The better you understand what is normal
and okay in the way of behavior,
what everybody's doing, where things are going,
the easier it's going to be to spot the anomaly.
And the anomaly is what you're interested in.
Give us a sense of an idea of what might be sticking out as unusual.
A ship who's conducting some nefarious activity might
do something a little odd.
Such as, it might be off on a piece
of the ocean that's not normally traveled,
because it's not in a direct line between different ports.
Why would the ship be there? Normally ships want
to take the shortest route to save fuel.
Or it might be right on track,
but for some peculiar reason in the middle of nowhere it slows down.
And normally ships won't do that.
But if you're transshipping something illegal
into a smaller vessel, for example,
and you don't particularly want to be seen,
you might choose to do it in the middle
of the ocean where you're away from prying eyes.
And to do that, you might need to slow down.
So what we want to look for is to understand what
is a normal pattern of shipping?
What are the normal tracks and speeds and kinds of behaviors?
And against that, do we see any ships
which are doing anything out of the ordinary?
One aspect of big data that the team at the Spetsia is looking at
is taking publicly available information
and crunching the numbers for NATO's benefit.
Let's say you have to set up an operation
to secure the waters of a country, for example.
Well, wouldn't it be nice when you go in,
when you're preparing, to know what the normal
patterns of life, patterns of maritime traffic are in that area,
the comings and goings of
the vessels and the ports.
So you also want to know sort of the layout of that maritime traffic.
Having that information available,
the historical information, really helps you eliminate some
of the haystack in your search for the needle in the haystack.
This tool helps you.
So to describe the haystack
and remove the parts that become known to you so you can
focus on really what is unknown to you.
Let's say you are interested in Syria.
We have data for three ports.
We want to go to this port here.
There's mostly cargoes, I mean way,
way more cargoes than any other type of vessel.
So perhaps that tells you something about the types of assets
and the types of operations
that you have to set up in that area.
That specific port.
Cargo, and then we can look at flags.
That's a typical question that operators have.
What are some of the things that you're
What are the vessels that have a specific state flag
that have visited a specific country?
Thank you.
These are very interesting, usually, null flags.
So these are vessels that have not declared their flags.
Now, if I leave that there, I'm highlighting.
Perhaps you can see it, perhaps not.
But there are three dots, one, two, three.
In some countries, there are quite a few nulls.
Vessels do not declare their flag and those tend to be
interesting vessels. That could be the needle.
It could be the needle, that's
correct. Where do you get that data from
and how do you get it into a usable state? We have
a feed from the US Coast Guard that's worldwide.
It's the AIS feed, it's the automatic identification system.
It's a broadcast system that vessels use.
And we gather all that data and
processing pipeline that puts it in a database
and then extracts it again from the database
and analyzes it and produces summary statistics
of vessels that have visited the various ports.
You can go directly to a map of the AIS,
so the Automatic Identification System, which
is the location that vessels have broadcast of themselves.
And now we were focused on Syria, so this tool brings up
Automatically, the vessel tracks of all the vessels
that have visited Syria in that month we had selected.
You can focus on specific types of vessels, for example,
passenger, which is an unusual one for Syria,
and yet there is some. You can see it here.
And again, you can hone in and look at the full data,
show all the columns,
and right there you have the full data set.
We want to make that information accessible
to the user in an easy way.
This is different from other approaches where you have
an algorithm that runs in the background
and surfaces just the last signal, perhaps an anomaly decision.
The problem with that is that what is an anomaly today
is not an anomaly tomorrow, and it wasn't
an anomaly the month before.
So these rule-based definitions of anomalies are a little brittle.
What this does, it doesn't really give you rules
for declaring anomalies, but it gives you information.
And the user can decide how to use that information best
and apply it to each individual analysis and use case.
How complicated, scientifically, is what we're looking at?
Actually, I'm glad you asked this,
because perhaps somebody would be tempted to say,
oh, you know, these algorithms are so complex,
only we can do it, you know, we have to do it.
But it's not true.
Actually, they're quite simple.
What it takes is attention to what the user needs.
And that feeds back into the design of the algorithms.
The algorithms, to answer your question specifically,
are data aggregation, filtering, and counting, essentially.
Because as you've seen, these are vessel counts.
This is unclassified, which is the other great thing about it.
It's open source. We don't give out this tool outside of NATO.
So, NATO nations already
have some advantages,
but for their vessels
at sea, one team at La
Spezia is adding a little muscle to a more direct problem.
So I'm in the rather curiously named Muscle Lab with Warren.
Warren, tell us why the Muscle Lab, first of all.
Where are we? What does it do?
The muscle systems we typically use
as our primary research platform for mine countermeasures work.
That's the focus of our group.
We're looking at the application
of robotics specifically to mine warfare.
These systems have become more and more robust over the years.
So now we're looking at how do we take what
is a very good automated platform,
and give it the decision-making and the evaluation capabilities
so that it can become a very effective platform for mine countermeasures.
MCM is typically organized around a series of phases.
Detection, classification, identification and neutralization.
And to date we don't really have one vehicle that can do all of those.
Maybe one vehicle is very good at identification,
maybe one is very good at detection.
What are some of the challenges that you face in terms of getting
the robots to communicate and to work together as a team.
A lot of times we'll have low bandwidth
and somewhat poor communications when you compare
it to what we're used to with Wi-Fi and mobile phones.
So we need to think about how
do they understand what each other is doing in potentially very
communications limited environments
and then how do they optimize themselves so that we
can still do all of those phases of MCM effectively.
One aspect that we are working a fair bit
on is how do we now integrate the robots with
With the manned fleets.
So the traditional mine hunter that most people have seen sort
of floating around in harbors,
those platforms have a set of doctrine
or a set of rules on how they do MCM.
So how do we adapt those rules for the robots?
Robots need to be able to make decisions
that are effective from a mission perspective.
But the humans also need to have trust in the platforms too,
as they're out doing the work.
So I think that's going to be the next big step,
is that decision-making and the effective mission processing.
So if that's what NATO's Centre
for Maritime Research and Experimentation
is doing right now, what about the future?
Are we talking about making systems like
the submarine obsolete in 30, 40 years?
Could it be that much of a game-changer for navies around the world?
It's going to be a real game changer,
but it's not going to make submarines obsolete.
That comes out of Marvel Comics
and James Bond movies rather than the world I think we really live in.
I think what we're going to see, though,
is some dramatic change in the way those systems are used
and the effectiveness of them
as they're leveraged by autonomous systems.
But for the foreseeable future, there'll still be a man in the loop,
...and there'll still be manned systems which are part of the whole...
security arrangement that you make for any situation.
So the Centre's been here for over 50 years,
will it be here in another 50?
I think it would be foolish to try and guarantee anything in 50 years,
but as a general principle
change is always happening, there
are always technological challenges and there will always
be a need for something like the Centre
to serve the nations for many years to come.
And that is it for this edition,
we hope you've enjoyed the programme, bye for now.

