A reconfigurable aerial–aquatic drone capable of operating in both environments through a mechanically actuated morphing architecture. Responsible for the complete mechanical design and CAD development.
Air and water differ in density by roughly three orders of magnitude, so the geometry that is efficient in one is close to the worst case in the other.
Aerial operation requires low mass and rotor geometry optimised for thrust in air. Underwater operation requires positive structural rigidity against hydrodynamic drag, sealed electronics, and thrust geometry suited to a far denser medium. A single fixed geometry satisfies neither condition well.
Current practice deploys two separate vehicles for missions spanning both environments — submerged pipeline inspection, or coastal survey above and below the surface — at the cost of two deployments and two operating crews. The design objective was a single airframe that reconfigures its geometry mechanically to suit each medium, rather than compromising between them.
Concept selection followed a literature review of morphing aerial vehicles and hybrid aerial–underwater platforms, alongside biological precedents for shape change at the air–water boundary. Two architectures were evaluated: a morphing fixed-wing platform and a reconfigurable multi-rotor.
The quadcopter architecture was selected. Published work identifies impact loading during high-speed water entry as an unresolved problem for fixed-wing configurations, whereas multi-rotor hybrids have demonstrated repeatable surface transition. With a single prototype build available, the lower-risk architecture was the appropriate choice.
Actuation. Reconfiguration is driven by two NEMA 17 stepper motors rated at 5.6 kg·cm, through DM542 microstepping drivers. Steppers were selected over servos because the mechanism must hold commanded position against continuous hydrodynamic load rather than simply reach it.
Sealing and shaft support. Every actuated joint represents a penetration through the sealed enclosure. Each drive shaft runs through a spring-loaded rotary shaft seal carried on needle-roller and stainless ball bearings, so a single interface provides both radial support and the pressure boundary. Stainless steel is specified throughout the wetted assembly: corrosion inside a sealed volume is neither inspectable nor serviceable.
Interactive 3D model — the full Fusion 360 assembly, 1,214 parts.
GLB · 28 MB · loads on demand1071 × 1021 × 173 mm across the rotors, and 1,214 discrete parts rather than a single merged shell — which is what makes it an assembly you can still work with rather than a picture of one. Nothing here is decorative geometry: the fasteners are the fasteners we bought, so the model doubles as the bill of materials and told us early where two of the morphing joints would foul.
The cycloidal reduction drive that actuates the morphing arms.
Interactive 3D model — the cycloidal drive assembly, 27 components.
GLB · 8 MB · loads on demandA cycloidal drive earns its place here because of what it does in a small space: a single eccentric input, a lobed disc rolling inside a ring of pins, and a large reduction in one stage — with load shared across many contact points rather than a couple of gear teeth. That matters for a joint that has to hold position against water drag without backdriving. The whole unit is 110 × 109 mm across and only 37.5 mm deep — 27 components, mostly off-the-shelf bearings and fasteners around the few pieces that had to be made: the eccentric, the cycloidal discs and the pin housing.
Waterproofing. Ingress protection was designed into the architecture rather than applied as a finishing operation: gasket sheet and O-rings at static joints, silicone sealing around the electronics bay, and rotary shaft seals at every dynamic penetration. Cable entries were sealed with adhesive-lined heat-shrink at each pass-through.
Manufacturing and assembly. Structural components were printed in PETG, selected over PLA for toughness and moisture resistance. Assembly proceeded in stages with alignment verified at each, since a sealed assembly makes early errors unreachable later.
Testing. Each motor was bench-tested for thrust individually before integration, so any imbalance would be identified as a component fault rather than diagnosed on the completed vehicle.
Capabilities developed through the design, build and validation of the prototype.
As built.