AUV vs ROV: What Actually Separates Them

The tether. That is the whole distinction, and nearly every other difference you will read about follows from it.

An ROV hangs off a cable that carries power and data up to a surface vessel, and a pilot flies it from a control van while watching the same camera feed you would. An AUV carries its own batteries, runs a pre-loaded mission, and nobody talks to it until it comes back. Two very different jobs fall out of that one choice.

What the tether buys you

Power, mostly. A work-class ROV draws between 100 and 250 horsepower down that umbilical, and it can draw it all day. The Saab Seaeye Leopard, a compact electric work-class vehicle rated to 3,000 metres, puts out more than 500 kgf of forward thrust from up to eleven thrusters while carrying a 100 kg tool payload. Schilling's UHD sits in the heavier category, rated to 4,000 metres as standard with payload capacity up to 300 kg.

None of that is possible on batteries. Not for a shift, anyway.

It also buys a human in the loop. When a pilot sees something unexpected, they stop, look, and decide. That matters enormously for intervention work: turning valves, cutting rope, changing out a subsea module, fitting a clamp on a damaged pipeline. Manipulator arms need judgement, and judgement still comes from a person sitting in a container on deck.

What the tether costs you

A vessel. That is the expensive part, and it is why the AUV market exists at all.

An ROV spread means a ship holding station above the vehicle for as long as the job runs, with a launch and recovery system, a winch, a control van and a crew working shifts. The tether limits range, so the ship follows. Day rates on the vessel dwarf the cost of the vehicle.

An AUV cuts the ship loose. Launch it, and it flies a survey line for hours without anyone watching. HII's REMUS 620 runs up to 110 hours on a charge with a 275 nautical mile range, dropping to 78 hours and 200 miles once you fit a synthetic aperture sonar and start drawing real power for the payload. Kongsberg's HUGIN family covers depth ratings of 1,000, 3,000, 4,500 and 6,000 metres, with the HUGIN Superior running to 6.6 metres long and 2,200 kg.

You give up reacting. An AUV on a survey line flies straight past something interesting, because nothing in the mission plan told it to care and there is no way to tell it mid-flight.

How the work splits in practice

Survey goes to AUVs. Mapping a cable route, running a pipeline inspection line, covering a wind farm site, building bathymetry over a large area: all of that is repetitive coverage of ground, and repetitive coverage of ground is what an autonomous vehicle is for. It flies a lawnmower pattern at a fixed altitude above the seabed and comes back with data.

Intervention goes to ROVs. Contact with infrastructure, decisions made at the point of work, force applied through a manipulator arm.

Inspection sits in the middle and gets argued about. An AUV can carry cameras and fly close enough to image a structure, but whether it does so safely without a pilot watching depends on the structure, the visibility, and how much you trust the obstacle avoidance that day.

The comparison table nobody writes down

Depth rating is not a useful axis of comparison, which surprises people. Both classes reach 6,000 metres. The REMUS 6000 does, HUGIN does, and plenty of work-class ROVs are built for full ocean depth on custom orders. Depth is a pressure housing problem, and pressure housing problems were solved decades ago. It is not the constraint.

The constraint is what happens when the plan meets the seabed.

Where the line is blurring

Resident vehicles muddy it considerably. Park an ROV in a subsea garage on a live field, run power and fibre to it from the platform, and fly it from an onshore control room. That vehicle is tethered, has full intervention capability, and needs no vessel at all. It breaks the tidy split described above, and it is being deployed now on producing fields in the North Sea and Gulf of Mexico.

Hover-capable AUVs push from the other side. They stop, hold position and inspect rather than only flying lines, which drags autonomous vehicles into work that used to need a pilot on a joystick.

If you are specifying for a job, ask two questions. Does anything need to be touched, and can a vessel be spared. Those two answers will pick the vehicle before any spec sheet does.