U.S. Marine Corps just tested a self-driving artillery platform

Key Points
  • Overland AI integrated its OverDrive autonomy stack and SPARK sensor kit onto the Marine Corps ROGUE Fires platform, demonstrated in April 2026.
  • The ROGUE Fires prototype operated autonomously over mixed terrain for several hours and conducted manned-unmanned teaming operations during the demonstration.

Overland AI has integrated its autonomous driving system onto the U.S. Marine Corps’ Remotely Operated Ground Unit for Expeditionary Fires platform, demonstrating the ability to maneuver an unmanned ground vehicle through contested terrain without human intervention.

The Seattle-based company announced the milestone on April 23, 2026, marking its first publicly confirmed work with the Marine Corps on an expeditionary fires mission.

The integration pairs Overland AI’s OverDrive autonomy stack with the company’s SPARK sensor and compute kit — together forming the hardware and software package that allows the ROGUE Fires platform to navigate independently across mixed terrain in environments where communications and GPS signals cannot be relied upon. At a recent demonstration, the ROGUE Fires prototype operated without human intervention over mixed terrain for several hours. The demonstration also included manned-unmanned teaming operations, with the ROGUE Fires platform coordinating with other vehicles during the exercise.

- ADVERTISEMENT - CONTINUE READING BELOW -

“Our work with the ROGUE Fires platform reinforces that Overland’s autonomy is capable of expeditionary operations inside the weapons engagement zone,” said Tim Francis, director of business development at Overland AI. “This technology is built to handle the contested radio frequency and climate demands of littoral missions, and we are proud to support the Marine Corps.”

The ROGUE Fires program is the Marine Corps’ effort to field a remotely operated ground vehicle capable of delivering fires — artillery, rockets, or missiles — without putting a crew at the point of engagement. The platform is designed specifically for expeditionary operations, meaning it needs to function in the kind of austere, rapidly changing environments the Marine Corps habitually operates in: littoral zones, island terrain, forward positions without established infrastructure. Removing the human crew from the vehicle removes the human cost from the most dangerous part of the fires mission — placing a weapons system close enough to engage targets inside the weapons engagement zone while keeping operators at a safe distance.

A vehicle operating inside the weapons engagement zone, in a communications-denied environment, without GPS, cannot wait for a remote operator to issue movement commands. It has to navigate on its own, respond to terrain obstacles in real time, and do it reliably across the kind of uneven, unpredictable ground that defines littoral and expeditionary environments. OverDrive was built specifically for that problem. Byron Boots, co-founder and CEO of Overland AI, was direct about the design intent: “Comms-denied and contested environments are not an edge case for the Marine Corps. OverDrive was built for those conditions.”

OverDrive functions as the brain of the autonomous vehicle — processing sensor data, building a real-time understanding of the surrounding terrain, and translating that understanding into movement decisions without requiring human input at each step. SPARK is the physical package that makes OverDrive work on a given platform: the sensors that feed it data and the computing hardware that runs the algorithms. Together they form an upfit kit that Overland AI can integrate onto existing military ground vehicles rather than requiring purpose-built autonomous platforms from scratch. That approach — adapting autonomy to existing hardware rather than designing new vehicles around it — is faster and cheaper than building from the ground up, and it allows the Marine Corps to evaluate autonomous capability on a platform already in development.

The multi-hour demonstration over mixed terrain without human intervention is a meaningful threshold. Sustained autonomous operation across varied ground — not a controlled track, not a single terrain type, but the kind of mixed surface that real expeditionary environments present — is significantly harder than short-duration demonstrations under controlled conditions. The addition of manned-unmanned teaming to the same demonstration signals that Overland AI is not just solving the navigation problem in isolation. It is building toward integrated operations where autonomous vehicles work alongside crewed ones, sharing information and coordinating movement in real time.

The company completed a month-long deployment with the 3rd Brigade Combat Team, 82nd Airborne Division, at the Joint Readiness Training Center — one of the U.S. Army’s premier combat training environments. During that deployment, Overland AI’s autonomous ground vehicles performed across logistics, reconnaissance, and counter-UAS missions under large-scale combat operations conditions. That validation, conducted with the Army at JRTC, established that OverDrive could perform across multiple mission types under realistic training pressure. The ROGUE Fires integration extends that record into a new service and a new mission set.

The broader context for both efforts is the U.S. military’s accelerating push to integrate autonomous ground systems into combat formations. The Army’s Transform in Contact initiative, the Marine Corps’ ongoing development of distributed and expeditionary concepts, and the Department of War’s growing investment in uncrewed platforms all point toward a force that uses autonomous vehicles not as experimental curiosities but as operational assets. Overland AI is positioning OverDrive as the autonomy layer that connects existing platforms to that future — a software and sensor package that can be integrated onto vehicles already in service or in development, rather than waiting for a new generation of purpose-built autonomous systems to complete full acquisition cycles.

For the Marine Corps specifically, the ROGUE Fires program addresses a problem that has defined expeditionary fires for decades: how to get lethal capability close enough to matter without putting people in the most dangerous position on the battlefield. Autonomous navigation in denied environments is a direct answer to that problem. A vehicle that can move itself, navigate without GPS, and operate without a communications link to a remote operator is a vehicle that can go places and do things a crewed system cannot — and survive conditions that would otherwise require accepting unacceptable human risk.

Readers who wish to follow our weekly coverage can subscribe to the Weekly Defense Roundup.

If you wish to report a grammatical or factual error in this article, please let us know by using the online form.

Executive Editor

Support The Defence Blog

Independent reporting takes resources. Join us on Patreon.

Become a patron

More Like This

U.S. Army’s top official tested laser-armed vehicle in New Mexico

The U.S. Army's top civilian official sat down at the operator's seat of a laser-armed pickup truck at White Sands Missile Range in New...

San Francisco startup’s hydrofoil boat wows U.S. Navy brass

A San Francisco-based maritime technology company's hydrofoiling electric boat stopped senior U.S. Navy admirals and captains in their tracks at the Sea-Air-Space conference, drawing...

Neros Technologies shrinks its attack drone controller by half

A Los Angeles-based drone technology company has redesigned its ground control station for FPV attack drones to fit on a soldier's body armor, cutting...

U.S. Army tests British-made interceptor to beat drones

The U.S. Army's 52nd Air Defense Artillery Brigade has tested a new low-cost interceptor called Skyhammer in Europe, putting Cambridge Aerospace's system through developmental...

U.S. Army invests $461M to rebuild short-range air defense fast

The U.S. Army is nearly doubling its investment in its primary short-range air defense system for fiscal year 2027, requesting $461 million for the...