The machine sits on the wet coping with water still running off it.
Four printed flippers, a clear tube of electronics, a camera the size of a thumb.
It weighs eleven pounds, roughly a large house cat.
What the robot turtle did on the way across the pool matters more than any of that.
It went from one end to the other without stirring the water, and that single property is the reason the thing exists at all.
What a flipper does that a propeller cannot
A propeller moves a vehicle by throwing water backward. The turbulence is not a side effect of the propulsion. It is the propulsion, and the noise comes with it.
Over a shipwreck that costs nothing.
Over a reef in trouble it costs a great deal, because the instrument starts damaging the thing it came to measure, lifting sediment, scattering animals and putting pressurized water across structures that break easily.
A turtle solves the same problem with a different motion. The front pair of limbs row it forward while the back pair steer, and the wake it leaves is something reef animals live with already.
That split is copied directly, two large flippers at the front for thrust and two smaller ones behind for stability and direction. The builder put the reasoning plainly.
He did not want, he told an American magazine, to harm the places he was hoping to protect.
What is inside the shell
The brain is a hobby computer the size of a credit card, the same board sold for classrooms and garage projects, sealed inside a clear acrylic tube. It runs the detection models, logs what it sees and sends it out wirelessly.
A location module lets it follow a preset grid on its own, which removes the tether that most survey machines drag behind them.
The sensing is more crowded than it looks. A camera in the nose, lights for dark water, an ultrasonic transducer that finds obstacles by sound, depth sensors, and a homemade holographic imager for particles too small to photograph.
Motorized fins handle buoyancy. A lithium battery gives about eight hours, a solar panel stretches that, and the whole thing cruises at half a mile an hour, which is not coincidentally the speed of a turtle.
The pool, and what the 96 percent measured
Almost all of the testing happened in a grandparents’ backyard pool a little over eight feet deep.
He printed a reef simulator, put it on the bottom and taught the machine what bleaching and invasive growth look like. Under those conditions it flagged replicated bleaching correctly 96 percent of the time.
Then it went into Lake Ontario, which is nothing like a pool. Murk, current and shifting light are what forced the front lights and the sound based obstacle finder onto the design, and none of that was in the original drawing.
The work took the national innovation prize at the Canadian science fair in Fredericton, then went to the European young scientists contest in Riga.
It came home with one of the four first prizes, worth around 8,000 dollars, from a field the organizers counted at 133 entrants representing 37 countries.
Where the pool stops resembling an ocean
The 96 percent deserves an asterisk, and the honest version of this story puts it there. That number came out of chlorinated water at a fixed depth under steady light, scoring printed models whose appearance the machine had been shown in advance. A reef is none of those things. It is cold, murky, moving, and its bleaching does not come off a printer.
Depth is capped by the acrylic housing and the pressure ratings of what is inside it, so the real ceiling is unknown until somebody goes and finds it.
Telling a real signal from a lookalike in open water is its own discipline, the same problem as a sensor learning to separate a whale from a drifting ghost net beneath a floating turbine. The next step is other environments, which is exactly what separates a prototype from an instrument.
What the price tag actually changes
The interesting thing about that computer is how ordinary it is. It is not specialist hardware and it does not want a support vessel, and reef monitoring has historically wanted both.
Everything else came off a home printer and a parts shelf, which puts the build inside a school budget rather than a research grant.
Cheap gear doing expensive work has a track record in this field, including the homemade torpedo that carried living coral 217 feet down so divers could start new colonies with it.
He wants to add water sampling next, he told the organizers, and the prize money is going toward it.
The pool is finished with. The ocean has not started.
Read the whole thing?
Get the week's signal, not the noise
Our sharpest reporting on energy, climate and nature — free, once a week.
