SandboxAQ Flies AQNav Magnetic Navigation on Northrop Grumman's Lumberjack, the First Reported MagNav Test on an Attritable One-Way-Attack Drone
Quentir Defense Monitor
Evidence-based insights for quantum defense and security. Published by Quentir Systems LLC · September 10, 2026.
On September 9, 2026, SandboxAQ announced the completion of a flight test that puts quantum-derived navigation on an affordable, attritable class of aircraft in the modern arsenal. The company's press release describes the campaign as the world's first reported test of a magnetic navigation system on an attritable platform, Northrop Grumman's Lumberjack, a Group 3 unmanned aircraft built for one-way-attack missions. It also claims a second first: the pairing of magnetic navigation with a visual navigation system on a one-way-attack platform. Both claims carry the company's own qualifier of first reported. Both point at the same underlying shift. Navigation that reads the Earth instead of listening to satellites has moved from experimental pallets aboard large transports to software riding a drone that is designed to never come home.
The operational background is plainly stated in the release itself. Small, expendable drones now operate routinely in airspace where GPS is jammed or spoofed, in Ukraine and around the Persian Gulf, and militaries are stockpiling them anyway. A munition that loses its satellite fix in the last kilometers becomes a statistic; the engagements that decide whether cheap mass works are fought exactly where the signal is worst. What SandboxAQ and Northrop Grumman tested is whether an answer to that problem can operate on the drone, with AQNav software installed into existing systems in under an hour.
This briefing reads the flight test as a defense capability: how magnetic navigation works and what was actually integrated, why the attritable platform matters more than another transport demonstration would, where the test sits among three independent quantum navigation trials reported in the span of a few weeks, and what still separates a September announcement from a navigation line item a program office can rely on.
How AQNav navigates: magnetometers read the crust's magnetic anomalies independently of satellite signals
The Earth's crust carries a pattern of magnetic anomalies, local deviations from the planetary field created by geology, that varies from place to place and changes only on geological timescales. Magnetic navigation, MagNav in program shorthand, measures that pattern in flight with a magnetometer and matches the measurements against survey maps to produce a position fix. The signal originates in rock and does not depend on a transmitted satellite signal, making the approach resistant to conventional GNSS jamming and spoofing. That independence distinguishes magnetic navigation from navigation aids that depend on receiving a transmission, although magnetic measurements remain subject to interference from fields generated by the platform itself. On its AQNav product page, SandboxAQ describes the system as passive and sensor-agnostic, pairing magnetometers with AI-driven signal processing that turns raw field measurements into positioning, and reports 9 platform integrations and 245 flights of accumulated validation.
The hard engineering problem is noise. A drone's own motors, actuators and electronics generate electromagnetic interference orders of magnitude coarser than the crustal signal being measured, and the announced test flew AQNav in its new software-first architecture, which is designed to filter that interference in real time using physics-based models and to operate with existing aircraft compute. That architecture is the practical news inside the announcement. AQNav now ships two ways: a full-stack version built natively into a platform's architecture, and a software-only version designed for compatibility with computers an aircraft already carries. On Lumberjack, engineers installed the AQNav software into existing systems in less than an hour. The public materials do not establish that sensor installation, calibration, configuration and complete integration all occurred within that hour, or explicitly confirm that the campaign required no added processing hardware. For a munition class measured in unit cost, limiting specialized navigation hardware can be a capability decision, and the same software-first approach can make the system portable across the drone fleets now being bought in quantity, provided suitable magnetometer and inertial inputs are available and configured.
The test also extends a flight history that has run through much larger aircraft. SandboxAQ has flown AQNav with the United States Air Force for more than three years, aboard C-17 Globemaster III and C-130J Super Hercules transports and through three large-scale military exercises, and has worked with Airbus and Boeing on commercial airframes since 2023. The system was selected for NATO's DIANA cohort in 2025 and participates in the Defense Innovation Unit's Transition of Quantum Sensing program, which evaluates MagNav for military autonomous systems. The trajectory across those entries is consistent: from demonstration pallets on transports, to a program office evaluation framework, to software on an expendable airframe.
Why the attritable platform is the point: a Group 3 one-way-attack drone imposes tight integration constraints
Northrop Grumman describes Lumberjack as a modular, AI-enabled unmanned aircraft system for joint operations and one-way-attack missions, positioned as affordable, attritable and scalable, and developed with partners Empirical Systems Aerospace and Palantir. The release adds that the design went from start to first flight in under 14 months. Attritable is a precise word in this market: the aircraft is intended to be affordable enough to accept its loss, which places cost, payload, power and integration constraints on every subsystem aboard. Expendable platforms have often relied on GPS as a precision navigation source available at little marginal cost to the aircraft. Jamming has weakened that advantage, and the documented density of interference across the Black Sea, the Baltic and the Middle East means designers must now assume denial as the normal condition.
That is the context in which a one-hour software installation matters. If GPS-denied navigation can be integrated with a drone's available sensors and computing architecture, then resilient guidance could become practical across a wider range of affordable platforms. The pairing with visual navigation on the same airframe reinforces the design philosophy: magnetic fixes work over open water, featureless terrain and at night, where cameras struggle, while visual navigation delivers fine precision where the scene is rich, and neither depends on satellite signals. What a force could gain, if mission-representative performance is established, is the ability to keep launching affordable aircraft into contested corridors with a better prospect of reaching their destinations. What the same technology could offer the defensive side of the ledger is identical navigation for resupply drones, transports and eventually civil aviation, which is why the readiness panel below reads the posture as it does.
Quantum pillar: sensing (inertial navigation and PNT). Use posture: dual-use. Technology readiness: This publication assesses the test at TRL 6 of 9. The AQNav software flew on the actual attritable platform it is meant to equip, a demonstration in a relevant flight environment, while the release publishes no accuracy figures, durations or routes from the campaign, so a buyer should read this as a successful integration trial one step short of a system proven in its final operational form.
Three alternative navigation results in a few weeks: Lumberjack's magnetic map, Q-CTRL's gravity fixes in the Coral Sea, and Saab's quantum-timed Giraffe radars
The Lumberjack test did not arrive alone. On August 27, 2026, Australia's Q-CTRL announced a maritime field trial in the Coral Sea in which its Ironstone Opal quantum gravimeter held roughly one nautical mile of positioning accuracy without any satellite fixes, surpassing navigation-grade GNSS backup systems by more than ten times, from equipment installed in an ordinary passenger cabin and running through heavy seas. This Monitor examined that trial in detail in its briefing on the Coral Sea demonstration. And in August, Saab UK reported a trial with Aquark Technologies, QinetiQ and the Royal Navy's Disruptive Capabilities and Technologies Office in which multiple Giraffe 1X radars tracked live targets while running on independent AQlock quantum timing sources without GNSS timing. Deliberately injected erroneous timing caused the unified air picture to degrade, and synchronization and the unified picture were recovered after correct timing was restored.
Read together, the three results describe a stack rather than a race. Magnetic anomaly matching gives aircraft a passive position reference, gravimetry gives ships one that works where the magnetic environment is unhelpful, and cold-atom timing keeps distributed sensors coherent when the shared satellite clock disappears. Each replaces a different dependency on the same vulnerable constellation, and each was demonstrated on operational or near-operational equipment rather than an optical bench, a pattern this publication has followed since quantum navigation first left the laboratory bench. For a planner assembling alternative PNT, the relevant observation is that the layers are maturing simultaneously in three different countries.
What separates a September announcement from a navigation capability a program office can buy
The release supports a precise claim: AQNav's software-first version was installed into existing systems on a Lumberjack in under an hour and flew successfully, paired with visual navigation, on an attritable one-way-attack platform for the first time anyone has reported. The public materials do not establish that sensor installation, calibration, configuration and complete integration all occurred within that hour, or explicitly confirm that no added processing hardware was used in this campaign. What the release does not support matters equally to a buyer. No accuracy figure, flight duration, route, altitude profile or magnetometer specification from this campaign is public, and the comparison that decides procurement, navigation error against a jammed GPS baseline across a mission-representative profile, has not been published for this platform class. Magnetic navigation also carries structural dependencies the announcement leaves implicit: fix quality tracks the resolution and currency of the underlying anomaly maps, performance varies with local crustal geology, and geomagnetic storms perturb the measured field.
Those are the questions the Defense Innovation Unit's evaluation framework exists to answer, and SandboxAQ's participation in the Transition of Quantum Sensing program means the answers are being generated against defense-relevant use cases rather than press timelines. A program office weighing this capability should ask for the campaign's error statistics, the map provenance and update cycle, and behavior during disturbed solar conditions. The direction of the evidence is nonetheless clear. Navigation that reads the planet's own field has now flown on the class of aircraft that is being procured by the thousand, with software installed into existing systems in under an hour. If mission-representative accuracy and reliability are established, forces that learn to field it could improve the prospect that their cheap mass still arrives on target after the satellites go quiet.
Sources
Primary source: SandboxAQ's September 9, 2026 press release via PR Newswire on the AQNav flight test aboard Northrop Grumman's Lumberjack, with remarks from Max Schuster of Northrop Grumman and Luca Ferrara of SandboxAQ. Other material: Northrop Grumman's Lumberjack platform page; SandboxAQ's AQNav documentation; Q-CTRL's August 27, 2026 Coral Sea announcement; Saab UK's quantum timing radar trial report.