Forterra Says Its Armored Autonomous ATVs Have Been Tested in Ukraine’s Deadliest Real-World Laboratory

Forterra, a U.S. defense technology company focused on autonomous vehicles, says more than 100 of its self-driving all-terrain vehicles have been operating in Ukraine for nine months, marking what it describes as the largest deployment of American autonomous ground vehicles in an active war zone. The company’s vehicles have been used for logistics and casualty evacuation under some of the harshest battlefield conditions in the world, offering a rare glimpse into how autonomy performs when GPS jamming, artillery fire and drone surveillance are part of the daily operating environment.

The deployment reflects a broader shift in modern warfare. In Ukraine, drones have transformed the skies into dangerous territory, pushing militaries to search for alternatives that can move supplies, transport wounded soldiers and support frontline units without exposing troops to the same risks. Forterra’s experience suggests that ground autonomy is moving from concept to combat test, but also that the technology still has a long way to go before it can operate independently in the chaos of war.

Why Ukraine has become a proving ground for military autonomy

Ukraine has emerged as the world’s most intense laboratory for military innovation. Since Russia’s full-scale invasion, both sides have adopted drones, electronic warfare tools, remote targeting systems and improvised robotics at a rapid pace. What began as a conflict that showcased the power of aerial drones has evolved into a broader contest over who can move safely across a battlefield where almost everything is visible to something else.

That reality has created demand for uncrewed ground vehicles, or UGVs, especially for logistics. If a road or track is under constant surveillance from the air, even a short supply run can become deadly. Ground systems that can move ammunition, food, fuel and medical supplies without placing drivers in direct danger can help reduce casualties and free soldiers for other tasks.

But the battlefield also exposes the weaknesses of autonomous systems. Mud, craters, broken terrain, damaged infrastructure and hostile electronic interference all complicate machine navigation. Unlike laboratory demonstrations, war does not give systems time to retry, recalculate or wait for a software update.

A battlefield shaped by drones and electronic warfare

One of the main reasons ground autonomy is attracting attention in Ukraine is the spread of first-person-view drones and other aerial platforms that can strike quickly and cheaply. According to U.S. Army autonomy program lead Sgt. Maj. Corey Wilkens, that aerial threat has made exposed movement far more dangerous than it once was.

“There’s nowhere to hide,” Wilkens said, describing a battlefield where soldiers can be attacked by FPV drones, munitions-dropping drones, artillery and mortars.

The combination of surveillance and strike capability means that any vehicle moving on the ground may be detected, tracked and targeted in minutes. In that environment, autonomous or remotely operated vehicles are not just a convenience. They can be a survival tool.

What Forterra deployed and how it was adapted for Ukraine

Forterra’s deployed platform is based on Polaris ATVs and fitted with the company’s own sensor and compute system. The vehicle, called Lancer, is gas-powered rather than battery-powered, a design choice that gives it a much larger operating envelope and greater payload capacity than many of the smaller Ukrainian-built systems already in use.

According to a Ukrainian soldier familiar with the platform, many locally built UGVs are optimized for short-range supply movement and typically rely on batteries. Those systems can be effective, but they tend to carry less cargo and often top out at around 250 kilograms. Forterra’s Lancer can carry roughly 750 kilograms, making it significantly more useful for heavier logistics tasks and medical evacuation support.

That difference matters on a battlefield where vehicles can be consumed quickly and every trip must justify its risk. The larger payload lets a single run move more material, which can reduce exposure and save time. For units under pressure, that can be the difference between a manageable resupply and a dangerous shortage.

Starlink and battlefield practicality

Forterra’s first versions were not immediately embraced by Ukrainian forces. The company’s systems were initially viewed as too tailored to higher-end American military requirements. One of the most important changes was adding a Starlink satellite internet antenna, which helped the vehicles fit more naturally into Ukraine’s operational environment.

That adaptation underscores a recurring theme in defense technology: the best system on paper may not be the best system in combat. A platform that works well inside a U.S. testing pipeline still has to survive in the mud, under jamming, with limited maintenance and with users who need reliable tools immediately rather than perfect ones later.

The numbers behind Forterra’s nine-month combat trial

Since arriving in Ukraine last October, Forterra says its vehicles have accumulated more than 2,500 miles of driving across more than 1,100 missions. The company says those operations involved moving 777,440 pounds of cargo and carrying out 52 casualty evacuations.

Those figures provide a rough picture of what autonomous ground vehicles can do when put to work continuously instead of being shown only in demonstrations. They also reveal the kind of support function these vehicles are filling. This is not a story, for now, about robot tanks or fully autonomous assault vehicles. It is about logistics, mobility and survival.

Some of the vehicles have been lost, especially when they became stuck in deep mud or were left vulnerable in terrain where Russian forces could attack them. That attrition is one of the major realities of combat testing: systems do not just fail technically, they fail tactically, often because the battlefield changes faster than the machine can adapt.

Key operational metrics

Teleoperation still dominates, and that is telling

For all the attention around autonomy, the systems are not yet operating entirely on their own in Ukraine. Ukrainian soldiers have mainly used them in teleoperated mode, meaning a human is still directing the vehicle remotely in or near the combat zone.

That approach reflects both prudence and the current state of the technology. The vehicles are valuable enough that losing them to a bad autonomous decision would be costly. Just as important, the software is not yet reliable enough to independently recognize unexpected enemy activity and respond appropriately in a live combat encounter.

A Ukrainian operator told TechCrunch that the system can handle navigation across varied terrain, but still cannot yet make the kind of real-time threat response required when an enemy appears unexpectedly.

In other words, the system can help move from point A to point B, but it cannot yet fully interpret the messy, ambiguous and adversarial realities that define modern combat. That gap is central to the future of military AI.

What Forterra says it learned from the battlefield

The company says the deployment has provided practical lessons that no testing ground could fully reproduce. Forterra Chief Growth Officer Scott Sanders, a former Marine officer, said the experience reinforces a wider truth about defense technology: the moment systems are used in real combat, hidden assumptions are exposed.

Sanders argued that no defense technology can be judged completely until it has been pushed into real combat conditions and confronted with the unpredictability of war.

Among the lessons Forterra says it has taken away are the effects of electronic warfare, the need to update software remotely, the importance of ruggedization, and the difficulty of keeping autonomous vehicles alive in environments full of mud, damage and hostile observation.

The company also learned where human labor is still unavoidable. Some steps in the workflow remain manual, including places where data must be re-entered or confirmed by hand. Scott Philips, Forterra’s chief innovation officer, said visiting a Ukrainian operations center made those bottlenecks visible in ways that a presentation never could.

Philips said seeing the system in Ukraine showed exactly where the process still breaks down, and where better software or automation could reduce pressure on soldiers doing the work in real time.

Why real-world testing matters for defense AI

For defense companies, battlefield feedback is valuable because it highlights the difference between “works in principle” and “works under fire.” Combat can reveal whether a vehicle can survive jamming, maintain connectivity, avoid getting trapped, and keep moving when the environment is constantly changing.

That kind of feedback is especially important for autonomy, which depends on data. Systems trained in benign environments often struggle when they encounter conditions that were never adequately represented in training sets. War presents an extreme version of that problem.

The technology challenge: combining robotics, software and AI

Forterra has been working on autonomous vehicles for two decades, long before the current boom in generative AI. The company is now trying to blend more established robotics techniques with newer forms of machine learning that could make machines more adaptable in unpredictable settings.

The goal is not simply to build a vehicle that follows a route. It is to create a platform that can better interpret changing conditions and respond without brittle pre-programmed behavior. But that is far harder than it sounds, especially when the vehicle must operate in an environment where no two missions are alike.

Sanders said some of the most important tasks are not well represented in public datasets or open-source models because they involve battlefield-specific behaviors humans do not normally perform. A system might need to move across a minefield, handle sensitive equipment or work around weapon systems in ways that standard AI models were never built to understand.

Sanders said the company has to combine classical robotics methods with AI, using each where it fits best rather than relying on a single approach.

That strategy mirrors a broader trend in autonomous systems: companies are increasingly treating AI as one component in a larger stack that also includes sensors, mapping, control logic, remote operation and human oversight.

Why open-source AI is not enough

In many consumer or enterprise applications, open-source models can help companies move quickly. In military autonomy, however, the mission profile is different. A battlefield vehicle must deal with novel threats, rough terrain and communication disruptions. It must also operate under strict reliability expectations.

That means the engineering challenge is not only about model quality. It is about systems integration, fallback logic and resilience under stress. A vehicle that performs well in a demo can still fail if it loses connectivity, cannot detect a new obstacle or makes a poor decision when data is incomplete.

Competitors are racing for the same defense market

Forterra is not alone in chasing the military ground autonomy market. Scout AI recently raised $100 million to develop foundation models and a broader family of autonomous systems for defense, including UGVs. Other startups, including Field AI and Overland AI, are testing similar platforms with the U.S. military.

The competition reflects a growing belief that ground autonomy is no longer a distant research project. Instead, it is becoming a procurement and deployment issue, with militaries eager to learn what works, what survives and what can be scaled.

That matters because defense contracting often rewards companies that can prove field utility before rivals do. A platform that survives combat and delivers measurable logistics value can gain a significant advantage in future procurement cycles.

The money and the market

Forterra says it has raised more than $500 million from investors including XYZ Venture Capital and Moore Strategic Partners. The company’s wartime experience could help it turn battlefield validation into future contracts with the U.S. military and allied forces.

In defense, operational evidence can be more persuasive than marketing. A company that can point to hundreds of missions in a live conflict has a stronger case than one that relies only on demonstrations or simulated exercises.

What military experts say ground autonomy can already do

Despite the current limitations, U.S. military autonomy advocates say the basic case for ground systems is no longer hypothetical. Wilkens said the field has already demonstrated that ground autonomy is possible today.

Wilkens said that ground autonomy is achievable now, and that the military has already seen proof of that in practice.

That is a meaningful shift in the conversation. The question is no longer whether autonomous ground vehicles can exist. It is what tasks they should perform, how much autonomy they should have and how quickly they can be trusted with more responsibility.

In the near term, the safest and most useful role appears to be logistics: moving supplies, hauling equipment, retrieving wounded personnel and reducing the number of exposed human movements in contested areas. Those tasks are less glamorous than combat robotics, but they are more immediately relevant.

The cost problem may decide how quickly the technology spreads

One of the clearest requests from Ukraine is that these vehicles need to be cheaper. Even if they are less expensive than some alternatives in their category, they are still valuable assets, which makes commanders cautious about where and how often they deploy them.

That caution stands in contrast to small aerial drones, which are cheap enough to be treated as consumables in many situations. Ground vehicles are larger, more complex and often more expensive to replace. As a result, every deployment must be weighed against the risk of losing the machine.

A Ukrainian soldier said attrition is unavoidable on this battlefield and that the unit had already lost several vehicles, underscoring the need for lower-cost systems that can be fielded more freely.

Lower prices could unlock more frequent use, more experimentation and faster iteration. But driving costs down while maintaining resilience, payload capacity and battlefield durability is one of the hardest problems in defense technology.

How this deployment changes the conversation in U.S. defense tech

Forterra’s deployment in Ukraine matters beyond the company itself. It provides one of the clearest examples yet of a U.S. autonomous ground vehicle being used at scale in active combat. That gives defense planners, investors and rival startups a much more concrete sense of what is possible now and what remains out of reach.

The larger lesson may be that autonomy in war is arriving in layers. First comes remote operation. Then comes constrained autonomy in navigation. Later may come more advanced decision support and, eventually, systems that can handle a larger share of mission logic independently. But each step has to be justified by performance under real conditions.

Ukraine has already shown how quickly military technology can evolve when a war creates urgent demand and constant feedback. Forterra’s vehicles are part of that story, even if they are not yet fully autonomous in the way futurists imagine.

Timeline: Forterra’s Ukraine deployment at a glance

What to watch next

The most important question is whether Forterra and its rivals can move from teleoperation to more trustworthy autonomy in live conflict. That will depend on better perception, stronger resilience against electronic warfare, improved software updates, and systems that can handle unforeseen battlefield events without human intervention.

It will also depend on economics. If these vehicles remain too expensive to lose, commanders will continue to treat them as precious tools rather than disposable battlefield assets. If costs fall, more units may be willing to use them in the same high-risk environments where drones are already common.

For now, Ukraine has given the U.S. defense tech sector a rare opportunity: a real war, real users and real consequences. Forterra’s vehicles are still far from the fully autonomous combat systems science fiction often promises. But they are already showing that ground autonomy has moved out of the lab and into the battlefield, where the real test is not whether the machine works once, but whether it can keep working when everything around it is trying to stop it.

That, perhaps more than anything, is why this deployment stands out. It is not a finished answer to the problem of military autonomy. It is a first hard chapter in the story.