A bounded error is not a position fix

For years the hard question at sea was not whether a ship could carry a clever sensor — it was what tells you where you are once the satellites are gone and your own instruments are quietly adding up their own mistakes. On 27 August 2026, a company in Sydney, Australia published an answer. Q-CTRL sailed a ship down 45 nautical miles of the Coral Sea and worked out its position with satellite navigation removed from every step of the calculation, by measuring the pull of gravity under the hull and matching that reading to a map of the planet’s own gravity. That is the new fact. It is not a satellite position fix. It is not a finished replacement for satellite navigation. And the news is not that the answer got small — it is that the error stopped growing.

What happened

On 27 August 2026, Q-CTRL, a company in Sydney, Australia, announced that it had navigated a ship with satellite navigation withheld from the entire position calculation. The technical manuscript went onto the open preprint archive arXiv the day before, on 26 August 2026. The sea trial itself was a week in the Coral Sea north of Cairns, Australia, in March 2026, aboard a 29-metre research vessel.

The navigation run covered 45 nautical miles, which is 83 kilometres, over about six hours. Satellite navigation was withheld from every stage of the calculation and reintroduced only afterwards, to score the result against a true track. Instead of a signal from the sky, the ship read the pull of gravity beneath the hull with a sensor built from a cloud of atoms chilled almost to a standstill, and matched the pattern it felt against a preloaded map of the planet’s gravity.

Here is the scoreboard. The plain inertial system — the kind that measures its own motion and adds it up, with no outside reference at all — drifted to about 14 nautical miles, or 26 kilometres, of error. The gravity-aided solution finished at 2.2 nautical miles, about 4 kilometres. More to the point, over the first 38 nautical miles, or 70 kilometres, the gravity-aided error stayed bounded at about 1 nautical mile while the plain inertial error kept climbing. One nautical mile is 1.852 kilometres, and nobody docks a ship on that. It is not a satellite position fix. It is something a plain inertial system cannot promise at all, because a machine that adds up its own motion also adds up its own mistakes.

The one idea

Gravity is slightly different in every place on Earth, and that pattern has already been mapped from space. Measure it precisely enough as you move and you know where you are — with nothing sent and nothing received, so there is nothing to jam and nothing to fake.

That is a capability class — working out where you are by reading the planet’s own gravity against a map, so position error stays bounded without any signal — not a company to name and not a product to buy. Learn the class. Leave the local use open.

Honest limits

• It is a preprint, not peer-reviewed, and all twenty-two authors work at the company.
• The company says the system beat a navigation-grade backup by more than 10 times; the paper’s own endpoint figure is about 6.3 times. The larger number is the company’s framing, the smaller one is the manuscript’s.
• One run, about six hours, 83 kilometres. Multi-week operation is named as future work.
• It needs a gravity map loaded in advance, and featureless seafloor gives little to match against.
• The package is still big: roughly a 30-litre sensor head plus a rack of lasers and electronics.
• Only the navigation branch excluded satellites; the fine survey numbers used satellite positioning.
• Gravity-aided navigation is decades old — the United States Navy ran trials aboard the USS Memphis in 1998 and 1999 with classical instruments — and measuring gravity at sea with chilled atoms is not new either. The narrow new thing is the first openly published, fully satellite-independent gravity-map-matching navigation run using a quantum gravity sensor.
• Magnetic-anomaly, terrain, sonar and celestial referencing are live alternatives on different paths.
• Naming a company, or “buy a quantum sensor”, is not the invention.

Sources

• Q-CTRL press release, 27 August 2026: https://q-ctrl.com/blog/q-ctrl-achieves-worlds-first-gps-free-quantum-gravimetric-navigation-demonstration-in-maritime-field-trial
• Everitt and colleagues, arXiv:2608.25563, submitted 26 August 2026: https://arxiv.org/abs/2608.25563
• Quantum Computing Report, 27 August 2026: https://quantumcomputingreport.com/q-ctrl-demonstrates-first-gps-free-quantum-gravimetric-maritime-navigation-in-coral-sea-field-trial/

Try inventing

Warmer Sun is where people come to follow, understand, and learn — not a lab. If you can restate the difference — an error that stops growing versus a position fix you can dock on — you understand the week.

You can try inventing with this at https://warmersun.com/forge/?q=spotlight-quantum-thornbay-gravity-fix-2026. Free signup. No card.