The ocean, unmanned

Right now, several thousand Argo floats are drifting through the world's oceans. Each one sinks to depth, drifts for days, profiles temperature and salinity on the way back up, phones home by satellite, and sinks again — for years, with no ship anywhere near. Most of what we know about the ocean's heat, and so about the climate, comes from this quiet robotic fleet.

On the surface, uncrewed wind-driven vehicles have sailed through hurricanes on purpose, streaming data from inside weather no crewed ship would survive. At the other extreme, researchers have flown a soft-bodied robot — silicone, rays-inspired, electronics spread thin so pressure has nothing to crush — in the Mariana Trench itself, more than 10,000 metres down.

And in 2022 the long game paid off in the Weddell Sea. Shackleton's Endurance, crushed by pack ice in 1915, lay under some of the worst sea ice on Earth. The Endurance22 expedition flew Sabertooth hybrid AUVs beneath that ice from an icebreaker, surveying lawn-stripes of seafloor at 3,000 metres until the ship appeared on sonar — upright, preserved by the cold, her name still legible across the stern. One hundred and seven years of hiding ended by a robot flying patiently under ice.

That is the century we are in: machines that find, see, reach, and preserve — which is precisely the work this Observatory exists to teach.

Endurance is an energy problem

Every long-lived ocean robot is an exercise in bookkeeping. A battery holds only so many watt-hours. Thrusters burn them fastest; sonar and lights next; computers always. The craft that stay down longest cheat: Argo floats barely propel at all (they ride buoyancy), gliders porpoise on wings, sail drones take their push from the wind.

The mission design question is always the same triangle: speed, sensing, and duration — pick two. A fast AUV with every sensor lit lasts hours. A float that mostly drifts lasts years. In between sits every real mission profile ever flown.

Where are the bottlenecks today? FIND is bandwidth-starved: mapping the other three-quarters of the seafloor is a coverage problem. SEE is power-starved: lights and cameras cost too much energy to run for long. REACH is materials-starved: pressure still dictates what can be built. PRESERVE is policy-starved: we can locate graves faster than we can agree how to guard them. Each bottleneck is a career. But before any robot can find anything, it has to move — next week, how machines swim.

Budget a 10 kWh AUV across speed, sonar, and lights. The planner shows how long the craft lives and how much seafloor it covers — then try to beat the survey target.