Aquatic Robots: Types, Uses, and How They Work

Aquatic robots are untethered or remotely guided machines built to operate in water, and they have evolved well beyond the torpedo-shaped submersibles most people picture. Today’s designs include fish-shaped swimmers with flexible tails, gliders that drift on buoyancy changes alone, soft-bodied grippers gentle enough to pick up a live jellyfish, crawling machines that walk the seafloor like crabs, and drone-like vehicles that transition between air and water. Their uses span ocean science, offshore industry, military surveillance, conservation, and even early planning for missions to ice-covered moons. What ties them together is a shared set of engineering puzzles: how to move efficiently underwater, communicate without radio waves, survive crushing pressure, and keep working for weeks without a recharge.

How Aquatic Robots Move Through Water

Propulsion is the first design choice, and it shapes nearly everything else about the vehicle. The most familiar approach uses thrusters or propellers, the same basic idea behind a motorboat. Thruster-driven autonomous underwater vehicles (AUVs) are fast and maneuverable, but they burn through battery power quickly, which limits mission length. For tasks that demand endurance over speed, engineers have developed alternatives that trade thrust for efficiency.

Buoyancy-driven gliders are probably the most energy-frugal option. Instead of spinning a propeller, a glider shifts a small volume of fluid between an internal reservoir and an external bladder. When the bladder fills, the vehicle becomes slightly more buoyant and rises; when the fluid is pulled back inside, it sinks. Wings convert that vertical motion into a gentle forward glide. The buoyancy engine developed for the Slocum glider, for instance, fits inside a hull just 20 centimeters in diameter and is rated for depths down to 1,000 meters, with a maximum buoyancy change of about one liter of displaced volume.¹ Lab tests on a smaller-scale glider showed that increasing the displaced volume from about 60 percent to 80 percent of capacity boosted average speed by roughly a third and buoyancy efficiency by over 40 percent.² These gliders can stay at sea for weeks on a single battery charge, collecting data across huge ocean transects with almost no noise or wake.

Biomimetic robots take a different path, copying the body plans of fish, rays, or eels. A fish-shaped robot propels itself by oscillating a tail fin, and the shape of that fin matters a lot. Research on a deformable caudal fin found that switching the fin into a rigid “instant” mode during the power stroke boosted thrust by about 27.5 percent compared with a conventional non-deformable fin, while a smoother sinusoidal deformation pattern still delivered an 18 percent gain.³ The advantage of biomimetic swimming goes beyond efficiency: a robot that looks and moves like a fish can approach marine wildlife without frightening it, which is useful for ecological surveys.

For work on the seafloor itself, some robots walk rather than swim. One design, called SILVER2, mimics benthic animals like crabs and octopi. It traverses irregular terrain, interacts delicately with the environment, and can hold position passively and silently, something a thruster-driven vehicle struggles with because it constantly blows sediment around.⁴ Another crawling robot uses hydraulic soft actuators with rigid reinforcements to achieve a remarkable payload-to-weight ratio of 5 to 1 while retaining the compliance to absorb impacts on rocky substrate.⁵

And then there are trans-medium vehicles that operate in air, on the surface, and underwater. The Loon Copter, for example, is a quadcopter with an active buoyancy-control system that lets it fly to a site, land on the water, flood a chamber to submerge, and then resurface and fly home.⁶ Managing the transition between air and water is the hard part: the propulsion system optimized for air drag performs poorly against the much higher resistance of water, and vice versa. These hybrid designs are still mostly experimental, but they hint at future robots that can survey a coastline from the air and then dive to inspect a pipeline in a single sortie.

Talking and Navigating Without GPS

Radio waves, the backbone of almost all terrestrial wireless communication, die out within meters of entering seawater. That single fact creates two of the toughest challenges in underwater robotics: how do you know where you are, and how do you send data back?

For navigation, most AUVs rely on inertial measurement units, which track acceleration and rotation to estimate position. The problem is that small errors accumulate over time, a phenomenon called drift. Terrain-aided navigation helps correct drift by matching sonar readings of the seafloor profile against a stored bathymetric map, much like a hiker cross-referencing a topo map with the ridgeline in view. One approach fuses terrain-based position estimates with velocity measurements from a Doppler sonar, using a simplified sensor setup based on a single Doppler velocity logger.⁷ Under-ice environments are even harder because there is no option to surface for a GPS fix; research on under-ice AUV navigation has focused on acoustic beacons and upward-looking sonar to map the ice canopy overhead.⁸

Communication underwater has traditionally meant acoustic modems, which send data as sound pulses. Acoustic signals travel far but slowly, and bandwidth is tight. A shallow-water experiment over a 3-kilometer range in the 10 to 32 kHz band achieved data rates of up to 60 kilobits per second using wideband signal processing.⁹ That is adequate for telemetry and short commands but far too slow for streaming video or transferring large datasets.

Optical communication is the emerging alternative for high-bandwidth links. Seawater is relatively transparent to blue-green wavelengths between about 450 and 550 nanometers, and engineers are now building laser-based systems that exploit that window. A dense wavelength-division multiplexing system demonstrated aggregate data rates exceeding 10 gigabits per second over a 20-meter underwater channel, and maintained performance even under increased water turbidity.¹⁰ A separate line of work on perovskite photodetectors has reached 20 megabits per second using a compact receiver tuned to blue-green light.¹¹ The trade-off is range: optical links work well over tens of meters but attenuate quickly beyond that, so they are best suited for close-range docking or robot-to-robot relay rather than long-distance communication.

Soft Robots and the Problem of Being Gentle

Many of the most scientifically valuable organisms in the deep sea are fragile: soft corals, gelatinous zooplankton, sponges, sea cucumbers. Traditional robotic manipulators, designed for the oil industry, grip with metal claws that can crush a specimen before a biologist even gets a look at it. Soft robotics offers a fundamentally different approach by building actuators from compliant materials like silicone elastomers and reinforced polymers.

The first use of soft robotic grippers in the deep sea targeted sampling on mesophotic reefs, the dim-light zone below conventional scuba range. Those early grippers demonstrated that compliant materials are inherently impedance-matched to natural environments, meaning they deform on contact rather than applying concentrated force.¹² A later design took the concept further with nanofiber-reinforced actuators capable of grasping live jellyfish, organisms so delicate that even gentle conventional contact can tear tissue. The gripper was tested successfully on three live jellyfish species in aquarium settings.¹³

One practical advantage of soft grippers is that they can be 3D-printed aboard a research vessel using inexpensive materials. During a deep-sea cruise, researchers fabricated custom soft manipulators on the ship and used them to collect a goniasterid sea star from rock at 1,473 meters depth, a specimen that would have been severely damaged by a rigid claw. They also demonstrated gentle pickup and release of a holothurian (sea cucumber) lying on sand.¹⁴ The ability to iterate on designs during a mission, adjusting finger shape or stiffness based on what you encountered on the last dive, is a significant practical benefit.

Surviving the Deep and the Cold

Pressure is the defining constraint of deep-ocean robotics. At the bottom of the Mariana Trench, roughly 11 kilometers down, the water column exerts about 1,100 atmospheres of pressure. Conventional robots handle this by sealing electronics inside thick-walled metal pressure housings, which adds weight, bulk, and cost. Soft robots offer a different strategy: because their bodies are made of nearly incompressible materials like silicone and oil, they can be directly exposed to ambient pressure without needing a rigid hull. The electronics are encapsulated in polymer, eliminating the air-filled cavities that would collapse under load.¹⁵

A striking example is a centimeter-scale morphable robot weighing just 16 grams that uses bistable chiral metamaterials paired with shape-memory alloys for actuation. At extreme depth, the increased stiffness imposed by hydrostatic pressure actually improves the snapping speed of its bistable units, turning what would normally be a problem into a performance advantage.¹⁶ This kind of pressure-adaptive design is still in early stages, but it points toward a future where deep-sea robots are small, cheap, and deployable in large numbers rather than the rare, multimillion-dollar vehicles they tend to be today.

What Aquatic Robots Actually Do

The applications fall into several broad categories, and some of the most impactful work is happening in environmental science rather than industry.

Autonomous eDNA sampling is a good example. Environmental DNA, the genetic material shed by organisms into surrounding water, can reveal what species are present in an area without anyone catching or even seeing them. Researchers coupled an Environmental Sample Processor (ESP) with an AUV and tested it in Monterey Bay. The system autonomously filtered water samples, preserved them for up to 21 days, and detected eDNA from taxa ranging from microbes and phytoplankton to krill and anchovy, with no significant differences in eDNA density compared to traditional manual sampling.¹⁷ A mobile platform like this can repeatedly sample the same water mass over hours, tracking how a biological community moves through space, something impossible with ship-based bucket sampling.

Coral reef conservation is another active area. Crown-of-thorns starfish outbreaks can devastate coral reefs, and manual eradication by divers is slow, expensive, and limited to shallow water. Researchers developed a remotely operated underwater robot fitted with an acetic acid injection device that can target individual starfish, delivering a lethal dose of vinegar solution while leaving surrounding coral unharmed.¹⁸ Looking ahead, a broader framework for reef management envisions combining AUVs, sensor networks, eDNA monitoring, and swarm robotics for scalable, adaptive reef protection.¹⁹

Offshore energy infrastructure also depends heavily on aquatic robots. Wind turbine foundations, subsea pipelines, and oil platform structures all need regular inspection for corrosion, cracks, and marine growth. Aerial drones inspect the above-water portions of turbine blades with high-resolution cameras and machine-learning algorithms that can classify surface damage with accuracy as high as 99 percent.²⁰ Below the waterline, ROVs and AUVs equipped with sonar and cameras handle the parts humans cannot easily reach.

Swarm Robotics Underwater

A single robot can survey a point. A swarm of robots can survey an area. Underwater swarm robotics draws inspiration from fish schools and insect colonies, where coordinated group behavior emerges from simple local rules rather than top-down commands. Decentralized control is especially well suited to the ocean because the communication constraints underwater make global coordination impractical; instead, each robot responds to its nearest neighbors, and useful collective patterns like formation-keeping, area coverage, and target encirclement arise without any central controller.²¹

The practical appeal is resilience. If one robot fails, the swarm continues. If communication is intermittent, individual robots can still operate using local sensor data and rejoin the group when contact is restored. The challenge, though, is that most swarm algorithms have been validated in simulation or in controlled pools. Scaling them to open-ocean conditions, where currents, turbidity, and acoustic noise are unpredictable, remains an active research frontier.

Keeping Robots Powered and Clean

Battery life is the perennial bottleneck. Most AUVs carry lithium-ion battery packs and must be recovered for recharging after hours or days, depending on the mission profile. One creative workaround is wave-powered propulsion: a vehicle designed to harvest energy from ocean surface waves achieved an average mechanical efficiency of about 45 percent across a range of wave conditions, with instantaneous power peaks near 68 watts under calm seas.²² A wave-powered AUV can theoretically stay deployed indefinitely as long as there are waves, which is most of the time in the open ocean.

Biofouling is the less glamorous but equally critical endurance problem. Within days of deployment, marine organisms begin colonizing any submerged surface, and within weeks a robot’s sensors can be covered in algae, barnacle larvae, and bacterial slime. This growth degrades optical sensors, clogs intakes, and increases drag. Anti-biofouling coatings, from copper-based paints to newer non-toxic polymer surfaces, are an active area of development, because long-term ocean monitoring depends on sensors that still work after months at sea.²³

Robots Headed for Other Worlds

The most ambitious application of aquatic robotics may not even be on Earth. Jupiter’s moon Europa and Saturn’s moon Enceladus are both believed to harbor liquid-water oceans beneath ice shells kilometers thick. Any mission to search for life in those oceans will need a robotic ice probe, called a cryobot, capable of melting or drilling through ice in temperatures as cold as 100 Kelvin, then transitioning to aquatic operation once it breaks through into liquid water.²⁴

The engineering challenges are staggering. The robot must function autonomously for months or years, with no possibility of real-time control from Earth due to signal delay. It must carry its own power source, likely a radioisotope system, and survive both the cryogenic ice transit and the unknown chemistry of a subsurface alien ocean. Laboratory melt probes have been tested in cryogenic ice at 79 Kelvin and warmer ice at 253 Kelvin to validate thermal models, but the gap between a lab bench and the surface of Europa is enormous. Developing the technology for such a mission is described as an “extremely demanding multistakeholder challenge” requiring collaboration across engineering, geoscience, and astrobiology.²⁵ In the meantime, under-ice AUV missions in the Arctic and Antarctic serve as proving grounds, testing autonomy, navigation, and survival in the closest analog Earth can offer to an ice-covered alien sea.