American autonomous ground vehicles are being deployed in active combat in Ukraine, marking the first large-scale real-world test of self-driving military technology under fire. Forterra, a U.S. autonomous vehicle builder, revealed that more than 100 of its self-driving ATVs have been operating in conflict zones for the past nine months, what the company believes is the largest deployment of autonomous ground vehicles in combat by any U.S. defense tech company.

Why Is Ukraine Using American Autonomous Vehicles?

The Ukrainian military faces a unique problem: drones dominate the airspace, forcing soldiers into dangerous positions on the ground. Ukrainian strategists have turned to autonomous ground vehicles as a solution to reduce human exposure to aerial threats. Ukraine is already building its own uncrewed ground vehicles (UGVs), but they have significant limitations. These homegrown vehicles are battery-powered and can only carry up to 250 kilograms of cargo, making them less versatile for logistics operations.

Forterra's Lancer vehicles, based on Polaris ATVs and equipped with custom sensors and computing systems, offer a substantial upgrade. They are gas-powered and can carry 750 kilograms of cargo, making them far more useful for moving supplies, ammunition, and evacuating wounded soldiers. Since arriving in Ukraine last October, the vehicles have driven more than 2,500 miles across more than 1,100 missions, carried 777,440 pounds of total weight, and completed 88 casualty evacuations.

"The bottom line is that this UGV for logistics and just maintaining our defense is the most important UGV in Ukraine. It's fucking fantastic, and we are dying to get more," said a Ukrainian soldier who has worked with the vehicles.

Ukrainian soldier, Ukrainian Armed Forces

What Are the Real-World Limitations of Autonomous Vehicles in Combat?

The deployment has revealed a critical gap between autonomous vehicle capabilities in controlled environments and the demands of active warfare. For now, Ukrainian soldiers are primarily teleoperating the vehicles in combat zones rather than letting them run fully autonomously. This human control is necessary for two key reasons: the vehicles are too valuable to risk losing, and autonomous systems simply aren't ready to identify and respond to enemy threats in real time.

The vehicles can navigate autonomously across diverse terrain, but they cannot yet identify unexpected enemy forces and react appropriately. This limitation highlights a fundamental challenge in military autonomy that differs sharply from civilian self-driving car development. A Ukrainian soldier explained the core problem: "We actually need to be able to respond to the enemy threats, live, while it's in front of the enemy, which the autonomy doesn't know how to do yet".

How Are Engineers Bridging the Autonomy Gap?

Forterra is working to combine traditional self-driving car algorithms with newer generative artificial intelligence (AI) software that allows machines to react to their surroundings in more flexible ways. The company began work on autonomous vehicles 20 years ago and is now applying lessons learned from the Ukrainian deployment to improve its systems.

One of the biggest obstacles is gathering the right training data. Military applications require data that simply doesn't exist in public datasets because they involve scenarios humans don't typically encounter. Scott Sanders, Forterra's chief growth officer and a former U.S. Marine officer, explained the challenge: "There's a lot of things you have to do that aren't available in an open source model because they're not things that humans do, whether that's figuring out how to navigate a minefield or operating a weapon system. You need to be able to turn the dials and some things more of a classical robotics approach, and then leverage AI where you need to".

Scott Sanders, Forterra's chief growth officer and a former U

"I believe this to be true of every defense technology that's ever been created,until you hit the realities of combat, you're just not going to know," said Scott Sanders, chief growth officer at Forterra.

Scott Sanders, Chief Growth Officer at Forterra

Steps to Improve Autonomous Military Vehicles Based on Combat Feedback

• Electronic Warfare Hardening: Forterra has learned critical lessons about protecting autonomous systems from electronic warfare attacks, a threat that doesn't exist in civilian autonomous vehicle testing.
• Remote Software Updates: The ability to update vehicle software from a distance has proven essential, allowing engineers to fix problems without physically accessing vehicles in active combat zones.
• Terrain Navigation Refinement: Real-world experience in challenging Ukrainian terrain has improved how vehicles maneuver in mud, snow, and uneven ground that typical testing facilities cannot replicate.
• Reliability Under Stress: Combat deployment has revealed which components fail under extreme conditions, allowing engineers to redesign systems for durability rather than just performance.

The Ukrainian military has also issued a clear challenge to Forterra: make the vehicles cheaper. While Forterra's Lancers benefit from Polaris' commercial supply chain, keeping costs down, they remain too expensive to deploy as freely as aerial drones. "Attrition is just a fact of this battlefield, and we have lost a few at this point, and it hurt, and we need more, and therefore we need them cheaper," a Ukrainian soldier told TechCrunch.

Forterra is not alone in this space. Competitors like Scout AI, which raised $100 million earlier this year, are training foundation models and developing autonomous platforms for the military that include UGVs. Other startups like Field AI and Overland AI are also trialing UGVs with the U.S. military.

The deployment in Ukraine represents a watershed moment for autonomous vehicle technology. While civilian self-driving cars have dominated headlines, military applications are advancing rapidly and revealing insights that could reshape how autonomy is developed across industries. The real-world data being collected in combat zones is invaluable, and it's showing that the path to true autonomy requires far more than algorithms alone.