Unmanned Underwater System

Context

Underwater is an unusually unforgiving place to put a machine. Water gets into anything that is not sealed, and a single failed seal ends the run and often the electronics with it. Radio does not travel through water, so the comfortable assumption that you can always talk to your vehicle disappears. Buoyancy and trim decide how the vehicle sits before control ever gets a chance to act. And the operator frequently cannot see what the vehicle is doing.

Most of the difficulty in this project came from those conditions rather than from any single subsystem being hard.

Approach

The design was ordered by consequence of failure rather than by interest.

Sealing came first, because it is the failure that cannot be recovered from during a run. Buoyancy and trim came next, because a vehicle that is badly balanced spends its control authority fighting itself and has none left for the task. Only then does the control logic become the interesting problem — and it is a genuinely different problem from a ground vehicle, because the medium pushes back continuously, momentum carries the vehicle past its target, and there is no ground to stop against.

The last constraint is communication. Because a radio link is not available underwater, the way the vehicle is commanded and observed has to be settled early; it shapes the whole architecture rather than being chosen at the end.

What I built

A working unmanned underwater vehicle for the competition: the mechanical structure and sealed housing, the electronics and propulsion inside it, and the control and command path used to operate it.

Testing was where the project actually took shape. Nothing about an underwater vehicle can be verified dry. A build that looks finished on the bench is untested until it has been in water, and every water test costs setup, drying and rework time — which puts a firm limit on how many bad assumptions you can afford to carry into one.

The vehicle competed in the TEKNOFEST Unmanned Underwater Systems Competition and reached the finals.

What I learned

The main lesson was to respect the environment before optimising for the task. Every hour spent on sealing and trim paid for itself, and every hour spent on clever behaviour before those were solid was wasted the first time the vehicle went in the water.

The second was about testing cadence. When each test is expensive, the useful skill is not running more tests — it is deciding what a test is supposed to answer before you get the vehicle wet.

Note: this entry is a structural placeholder. The competition record is accurate; the technical write-up still needs the owner’s own detail.