Charting Terradepth's Big Ambitions in the Unmanned Vehicle Space

Ocean mapping at scale is the target of a new unmanned systems player started by two ex-US Navy SEALS. They have big ambitions, from new building unmanned vehicles to creating the intelligence that will drive them to changing how accessible ocean data is. Elaine Maslin learned more.  

The pace of change in the ocean mapping space has been fast in recent years. New entrants have come into the scene offering “force multiplier” mapping using autonomous underwater vehicles (AUVs) and unmanned surface vessels (USVs). Shell’s Ocean X-Prize challenged technology developers to come up with ever faster mapping techniques in deep water and to push new boundaries. And yet, largely, many operations are still being done the same way.

AUVs are now also a relatively mature technology, although companies are nibbling at alternative ways to deploy and recover them without the need for manned vessels. The latter is just what a new outfit, Terradepth, is working on – complete with a new design of AUV, with hardware build support from Canada’s Cellula Robotics. Founded in 2018, Terradepth wants to bring a new multi-AUV mapping concept to market. Crucially, it’s one that doesn’t rely on surface vessels. Terradepth is also looking to change the model around how underwater data is owned and sold.

The company, based in Austin, Texas, doesn’t come from a subsea or even hydrographic background. It was formed by two former US Navy SEALS, CEO Joe Wolfel and President Judson Kauffman, with a mission to increase seafloor mapping rates economically and to create a business based on the data it gathers.

Terradepth CEO Joe Wolfel.

An ocean data challenge

Their focus came from personal experience, specifically the collision of the USS San Francisco – a nuclear powered submarine – with a seamount at flank speed in 2005. While they were deployed in Iraq at the time, it was big new. “How does the US navy not know there's a mountain underwater? That idea stuck in our heads,” explains Wolfel. They did a little digging and discovered that 70% of the earth is ocean “and we really don't know anything about it. The ocean is drastically underexplored. So we want to create a data repository so we can change the information market.

“But we have a data acquisition problem because it's too expensive to collect deep ocean data today. A ship goes out to sea, takes an AUV, it goes out for 12 to 48, maybe 72 hours, depending on energy use and sensor payload; it could be doing bathymetry, physical imagery, it could be a side scan sonar looking for a snapshot. You pull the robot up, swap the batteries, get the data and put it back in the water. The cost of this is between US$50,000 and $250,000 a day, just to run this ship. It's a lot. And it’s not scalable. We can't get the data, so we're going to go get it with a fleet of autonomous hybrid vehicles (AxVs) that will remove the requirement for that surface ship that we see as the main cost driver.”

Not an AUV but an AxV

Wolfel says they looked at the current AUV market, at Hugins and Slocums, but decided to build their own with their own concept of operation, focused on deepwater, where they see their concept having market-changing potential. The result is a lithium-ion battery-based multi-AxV system with hydrogen fuel cell repowering with each vehicle acting as a receive and transmit node so it can operate both at the surface as a gateway to send data to shore – building on constrained communications environments the pair experienced in Iraq.

“The AxVs are powered by lithium-ion batteries which, given our heavy payload requirements and small size, must be recharged after each day of use,” explains Kauffman. “So, at the end of a day of subsea data collection, the AxV comes to the surface and it's twin dives down to pick up where it left off. The now surfaced vehicle establishes GPS signal with a satellite, establishes acoustic communication with and positioning to the submerged twin (which will also use its own INS and DVL for positioning), and turns on its generator to begin recharging its batteries. This process repeated itself until we're low on fuel.”

Kauffman says this could run for 30, maybe 60 days. They call it the leapfrog, named after for a basic tactical combat manoeuvre the pair learned as Navy SEALs). “This way we always have accurate positioning and SATCOM and we're constantly collecting data without the need for a surface support ship,” adds Wolfel. It's a bandwidth constrained environment, so processing and transmission via satellite will be done while each AxV is at the surface.

“The AxVs are powered by lithium-ion batteries which, given our heavy payload requirements and small size, must be recharged after each day of use,” said Terradepth President Judson Kauffman

Fail fast

Terradepth is aware of the underwater challenge and are taking a fail fast approach but with experts from various fields on their books, including Seagate Technologies, a data storage firm, which also led an $8 million investment round into the company last year. A lot of focus is on the system’s intelligence.

“A lot of what we're doing on the hardware side is not particularly innovative,” says Kauffman. “Where we’re innovating on the software side, around the brain of the submersible. It will know when it sees things like a shipwreck or a thermal vent and it will know if it needs to turn on a camera or a video camera or do a second pass or surface and contact headquarters here in Austin and say ‘I don’t know what this is what should I do’.”

It’s a big challenge. “Submarines have been able to go to bottom of the ocean and collect whatever data we want but we can’t do it at scale because it’s been impossible to build autonomous systems that can do that. It’s only in last five years that technology i.e. machine learning and artificial intelligence has come to the point where it’s even feasible to put a massive fleet of autonomous deepsea submersibles out there and have them map the ocean.”

The other side of the coin is cost. The only subs with autonomy are the likes of Boeing’s Echo Voyager. “But that’s different; it’s not scalable at US $120 million apiece. We’re building these for, we think, less than $2 million apiece,” says Kauffman.

Google Earth - underwater
Terradepth has hired people from Tesla, people who know autonomy and AI, software specialists and technologists from other industries to all put their minds to this. “That’s where you get disruption and that’s critical to make everyone’s job easier, because academics and researchers can’t get the ocean data they need,” says Kauffman. Sonar systems are already being tested in Lake Travis, Texas, and this has included gathering data to put into training machine learning systems. Indeed, AI and machine learning on sonar data is a big part of the company’s focus.

“Our robots will be able to automatically detect anomalies within the sonar data in real-time, while submerged, and make decisions around that target recognition without needed a human's input or oversight,” says Kauffman. Testing of a 9m-long, 1 m-deep prototype (the commercial system will be smaller) is due to start in Lake Travis and then the US Gulf of Mexico closer in October. Tank testing of the energy recharge system was due to start in August.

The ultimate goal is something like a Google Earth type software system users can sign into and then go to the bottom of the earth and fly through a 3D point cloud terrain we’ve built, says Kauffman. And they’re not just talking about a map. They’re looking at building multi-dimensional data set to include <1m bathymetry, but also water column data, temperature, salinity and other chemical properties, as well as organisms detected there, and how that compares with one or five years ago in the same area.

“Climate change is a real thing, but a lot of people think we know why or how – and we don’t because we don’t know what’s happening on our planet because it’s too expensive,” says Kauffman.  

Targeted mapping

The company is already looking at where needs mapping, something projects like The Nippon-Foundation-GEBCO Seabed 2030 project will help inform, based on conservation and commercial metrics. As for who will buy the data, Wolfel suggests governments, exclusive economic zone authorities, bodies like the US’ National Oceanic and Atmospheric Association, but also telecommunications firms and others. There’s also potentially salvage hunters. There’s $400 billion of sunken treasure we know about but don’t know where it is because it’s too expensive to go looking, says Kauffman.  

It’s not quite the carer Kauffman originally envisioned when he left the Navy – he’d fancied the music industry, but then went into focusing on human capital in Silicon Valley, including working for Seagate Technology. “In Silicon Valley, I learned about some impressive stuff happening with autonomous systems, robotics and VR, and one day I just thought ‘is anyone applying this to the undersea space’, and  took a high level look at it; Ocean Infinity, Sail Drone and so on were just getting off the ground.”

Now Terradepth has joined the pack and Kauffman and Wolfel haven’t looked back.
How a Terradepth UxV will look. Graphic from Terradepth.