Robots are being designed to go where no human can, and this includes the depths. But what will they look like?

The search for a form-fit-for-purpose has been going on for a long time. Back in the 80s, some AUVs started to develop a glider shape to gain range, making the most of the underwater thermals by adjusting buoyancy and wings.

Although nowhere near as fast as conventional AUVs, this idea extended ocean sampling missions from hours to weeks or months and in 2009 one, the RU-27, became the first to complete a transatlantic journey. The US Navy’s Liberdade class pointedly mimicked a well-known form: the ‘ray’ like body and wings achieve a high level of hydrodynamic efficiency.

However, more recent developments such as the lithium ion battery paired with automatic undersea charge points promise to change the game again. This time, it’s the turn of the traditionally box-shaped ROVs to wander into the field of biomimetics.

Take, for example, the serpent-like robot being developed by Eelume for underwater inspection and light maintenance: the robosnake will dock at a charge and data hub on the seabed allowing operations to be programmed from a landside control centre.

Kristin Ytterstad Pettersen of project partner NTNU explained that while there were “numerous alternative shapes”, the sinuous structure stood out as having distinct advantages.

“These robots should at least be able to traverse from one subsea structure to another... eels and snakes are hydrodynamic... so much more efficient at travel than a box-shaped ROV.”

Secondly, these robosnakes can be a modular build, with extra joints being added as required to give flexibility. “In fact in a way it’s more like a multi-jointed arm,” said Dr Pettersen, one that will move through as many degrees of freedom as necessary with tools either at the head or tail “or both”.

Further, adding more sections will increase the body’s capacity without increasing the diameter. This point is particularly important as the robosnake has been designed to glide into confined spaces.

Certainly its free swimming capability and low noise level might provide it with work in aquaculture installations where it could find a job monitoring the stocks and repairing nets, “especially as we think it might not frighten the fish as much as a large ROV”, she said.

But what about the oil and gas market? Here, the potential gains extend well beyond the need for a ‘resident’ work ROV to meet day-to-day demands. She explained that getting a traditional, tooled up ROV into all the possible maintenance areas has resulted in “subsea oil and gas installations being eight times as expensive and 12 times as large as corresponding installations on land”. In other words, devices like the robosnake development could result in subsea plants becoming both smaller and cheaper to run and build.


Some developers have taken it a step further: Ocean One has been designed to be humanoid. Or at least, 'mermaid-oid'.

Recently tested in the waters off Toulon, the prototype has two arms which end in three-fingered hands, high-definition cameras for eyes and is moved by small, motorised propellers. But why something so similar to a human?

The fact is, despite having its own ‘brain’ which coordinates stability, position and a number of other elements, Ocean One aims to be less robot and more avatar.

The idea is to allow scientists to explore the depths themselves. It’s been achieved by what the developers at Standford University call ‘haptic feedback’, similar in principle to the system used by surgeons in order to carry out remote operations. The forces sensed by the merbot's hands are mirrored on the controls above - it’s so sensitive the operators on dry land can ‘feel’ the weight and texture of an object.

In fact on its first outing, an exploration of the wreck of King Louis XIV’s flagship La Lune (organised in collaboration with DRASSM), the avatar was able to pick up a small vase and return it to the surface.


At the other end of the scale, science could mimic nature’s swarms with fast, light and above all replaceable individual elements.

Martin Ludvigsen, a researcher for both NTNU and founder of Blueye Robotics told MJ that “drones presently seen in the air will soon move underwater”. Up till now marine robotics has been exclusively a high cost and professional-only field, but drones within a domestic budget may be about to change the game – and Blueye itself is about to start producing one at a price point “somewhere around US$2,000”. Again, the form follows function; it has a hydrodynamic shape almost like an Angelfish, higher than wide to make it stable enough for the video function.

Importantly these will have a simple interface, enabling them to be controlled via iPad, laptop or virtual reality headset – and linking them up in a swarm scenario will help cover a bigger area for seabed mapping, or broadening missions – one with a camera, another with a sidescan for example. And some processes require a simultaneous presence, such as looking for fish or measuring water currents in a fjord. He added: “Even if they’re only a fraction as popular as aerial drones – they will help drive development”.


However, one single, stubborn, stumbling block remains an issue: radio signals don’t travel far through water, “and this poses a big problem for navigation and communication,” said Martin Ludvigsen.

There are the tethered vehicles – which have a cable link that allows at least some of the brains to be up on the surface but there’s a trade-off: movement is restricted. Ocean One’s ability to convey sensory experience up to the surface demands enormous data bandwidths, so the merbot has a cable attached to a ‘companion’ which provides a robust surface connection. Although Ocean One might soon slip the leash thanks to a wireless link, it will still have to stay in a circumscribed area near its ‘friend’.

Even Blueye’s new drone will have a physical link, though Mr Ludvigsen explains that it’s only the thinnest of fibreoptics as many ROVs “use most of their energy just pulling a cable around”.

On the other hand, if the need is for a range greater than a few hundred metres, then the design is driven into an untethered operation. However, this requires keeping both ‘brains’ and ‘brawn’ - the energy source – onboard, and it can be a tricky balance as extra weight means more power.

Despite the challenges “we are working at removing the tethers, to enable the vessels to move freely”, said Mr Ludvigsen. Early developments used acoustic signals, “but these travel really quite slowly and the range is fairly limited to a small bandwidth” he added. However, “these last few years there have been more developments with underwater lasers which have a higher bandwidth”. It’s the kind of connection that sophisticated ROVs such as Ocean One might soon be using, but there are shortcomings: these laser links have a lower range – up to 100m – and even this is very dependent on water clarity.

Still, Mr Ludvigsen is confident the issues will be solved by a merging of “the megatrends of autonomy, artificial intelligence, machine-learning and computer vision”.

By Stevie Knight