A drone that lands reliably on a concrete pad can fail completely on a boat. The pad is not level, it is not still, and it is not stationary — it pitches, rolls, and heaves while translating at 10 or 15 knots, surrounded by a steel superstructure that distorts the compass and a sea that offers no second chance if the aircraft misses. Maritime operation is the environment that exposes every soft assumption in a drone design at once.
The demand is real: offshore energy inspection, fisheries and law enforcement patrol, ship-to-shore delivery, and search and rescue all want an aircraft that launches from the vessel it works with. Here is what that requires.
The Deck Is Moving in Six Directions
Three of a ship's six degrees of freedom dominate the landing problem: heave, roll, and pitch. On a small vessel in a moderate sea a deck can heave several feet and roll well past 10 degrees, with peaks that are not evenly spaced, while translating at the vessel's speed over ground.
That leads to the central design decision: land relative to the deck, not relative to the earth. A landing that targets a fixed latitude and longitude is useless on a moving vessel. The aircraft must know where the deck is with respect to itself, continuously, through the last 30 feet. The usual approaches:
- Visual fiducial markers on the deck, tracked by a downward camera. Cheap and effective in daylight; degraded by spray, glare, and darkness unless actively lit or infrared.
- RTK GPS on both aircraft and vessel, differenced to produce relative position. Good accuracy, but dependent on a link and on both receivers holding a fix.
- Ultra-wideband or optical beacons on the deck, giving relative range and bearing independent of satellites.
- Deck motion prediction. An IMU on the vessel predicts when the deck will be near the middle of its heave cycle, and the aircraft commits to touchdown in that window rather than chasing the surface.
Most working systems combine two, with the marker or beacon carrying the last few feet. The foundations are in precision landing and altitude sensing; the sea-state case adds the prediction layer on top.
One subtlety: a radar or laser altimeter over water reads the sea, which is itself moving, not the deck. The altitude reference has to switch cleanly from sea to deck during approach, and that handover is a classic place for a hard landing.
A Steel Ship Is a Bad Place for a Compass
Drone navigation leans on a magnetometer for heading. Park an aircraft on a steel deck beside a crane, a container, and a running generator and the magnetometer reads the ship, not the earth. The result is a toilet-bowl departure or a wrong-way flyaway seconds after takeoff — one of the most common ship-borne drone failures.
Mitigations, in order of effectiveness:
- Do not use the magnetometer for heading on deck. Dual-antenna GNSS heading gives true heading with no magnetic dependence at all, and is the standard answer for serious maritime aircraft.
- Initialize away from steel. A launch position on a boom, a mast platform, or the highest clear deck cuts distortion substantially.
- Visual-inertial odometry for the near field, which ignores magnetic fields entirely — the toolkit in GPS-denied drone navigation.
GPS itself is usually fine offshore, but multipath off wet steel is real, and reception in a superstructure's shadow can drop enough to matter on final approach.
Capture Systems: Nets, Hooks, and Deck Locks
For anything larger than a small multirotor, or for sea states above a gentle swell, free landing is replaced by capture. The aircraft flies into a mechanism instead of settling onto a surface.
| Capture method | Airframe suited to it | Deck footprint | Main risk |
|---|---|---|---|
| Net or barrier | Fixed-wing, small VTOL | Large, needs clear approach | Airframe damage on capture; net rigging |
| Wire / sky-hook snag | Fixed-wing | Small, vertical | Precise approach required |
| Deck grid + aircraft harpoon or claw | Heavier VTOL, helicopter-type | Moderate | Mechanism reliability, deck fitting |
| Enclosed box / garage | Multirotor | Moderate, self-contained | Alignment tolerance in swell |
| Free deck landing + tie-down | Small multirotor | Small | Only works in calm conditions |
Whatever the method, the aircraft must be secured within a second or two of contact. An unsecured drone on a rolling deck slides and goes over the side, and deck crew reaching for spinning propellers is its own hazard. Automatic locking — magnets, a latching cradle, spring jaws on the skids — is worth the mass. The enclosed version is the marine form of a drone-in-a-box docking station, with the added twist that the box is moving.
The Air Over the Water Is Not Calm
Two aerodynamic effects surprise teams that developed over land. First, a vessel's superstructure creates a turbulent wake extending well aft of the ship, with sharp vertical gusts exactly where the approach path lies, producing sudden sink and roll upsets a few feet above the deck. The answer is a defined approach sector relative to the ship's heading and the relative wind, enforced by the flight software rather than left to the pilot.
Second, downwash over water gives no clean ground effect: the rotor blows spray up into its own inflow, so at low hover heights the aircraft flies through a cloud of salt water it is generating. That is both a control problem and a corrosion problem, and a reason to minimize low hovering over the sea.
Salt Never Stops Working
Everything on a maritime aircraft corrodes faster than the same part ashore, because salt aerosol deposits, absorbs moisture, and forms an electrolyte that stays active long after the aircraft dries. The design responses are conventional but non-negotiable: sealed connectors with gold contacts, conformal-coated boards, stainless fasteners isolated from aluminum structure, sealed motor bearings, and drain paths so no cavity holds brine — see corrosion protection for metal parts.
The operational half matters as much: a documented fresh-water rinse after every flight, and a design that makes the rinse easy. Crews skip anything that takes ten minutes on a working deck.
When It Goes in the Water
Plan for ditching, because it will happen. Three questions decide whether a ditched aircraft is an incident or a loss:
- Does it float? Closed-cell foam in the arms or sealed fuselage volume costs almost nothing and turns a sunk aircraft into a recoverable one. Aim for positive buoyancy with the payload attached.
- Can you find it? A floating aircraft in a moderate swell is invisible from 100 yards. A bright float, a strobe that survives immersion, and an independent locator beacon with its own battery make recovery realistic.
- Is it safe to handle? Lithium packs that have been in salt water are a fire risk: disconnect, isolate in a metal container, do not recharge. Say so in the manual.
A parachute helps here too — a controlled descent into water beats an uncontrolled one, and the same system serves the flight-over-people case in parachute recovery systems.
Staging the Development
Maritime capability is added in layers, and skipping layers is expensive. Prove the aircraft on land. Then prove relative landing on a moving truck bed, which gives translation without swell. Then a moored vessel with wave-induced motion, then a vessel underway in calm water, then increasing sea state with a recovery boat standing by. Regulatory work runs in parallel: operating from a moving vehicle and flying beyond visual line of sight both need attention under the rules in FAA Part 107 certification.
Projects House develops drones and marine hardware for environments that punish shortcuts — relative-navigation landing, capture mechanisms, sealed and corrosion-resistant airframes, and the staged test program that proves each layer before the next one. Tell us about the vessel and the mission through our contact form.