Ground Systems — Updated March 2026

AI COMBAT ROBOTS &
HUMANOID SOLDIERS

The convergence of robotics, autonomy, and AI is putting armed machines on the battlefield. From armed quadrupeds patrolling USAF bases to humanoid robots breaching doors in Ukraine — this is the next front of AI warfare.

12+ Active Systems
$16B Market by 2030
6 Deployed Globally
2026 Combat Testing Active
TIME MAGAZINE — March 9, 2026: "The Race to Build AI Humanoid Soldiers" — front cover story confirming the field has reached mainstream inflection.

The Convergence of Robotics + AI for Ground Combat

For decades, autonomous ground systems remained relegated to laboratory demonstrations and limited EOD roles. That era is over. The simultaneous maturation of large-scale machine learning, low-cost actuators, edge computing, and battlefield networking has created the conditions for a genuine revolution in ground warfare — one driven not by missiles or aircraft, but by robots that walk, crawl, roll, and now shoot.

The modern AI combat robot is defined by three converging capabilities: perception (sensors + computer vision to understand the environment), mobility (quadruped legs, tracked drives, or bipedal locomotion to navigate terrain), and lethality or utility (weapons mounts, ISR payloads, or logistics functions). When all three mature simultaneously — as they are now — the result is a system that can operate where it would be too dangerous, too slow, or too expensive to send a human soldier.

The Ukraine war has compressed what would have been a decade of testing into two years of real-world deployment. Both sides are fielding ground robots with increasing autonomy, using the conflict as a live laboratory for doctrine, counter-tactics, and system refinement. What emerges from that crucible will define the infantry combat of the 2030s.

Combat Robots & Autonomous Ground Systems

01 Deployed
Ghost Robotics Vision 60
USA — Ghost Robotics / SWORD Defense
The Vision 60 is the most operationally significant armed quadruped in the world. Deployed by the USAF and USMC for perimeter security and base patrols, it carries the SWORD SPUR (Special Purpose Unmanned Rifle) remote weapon station — capable of firing a standard 6.5mm Creedmoor rifle. Its four-legged gait allows it to traverse terrain impassable to wheeled or tracked vehicles, and it can operate autonomously or under remote control. Current deployments include Tyndall AFB and overseas installations.
Quadruped Armed USAF / USMC 6.5mm Creedmoor Perimeter Patrol
02 Testing
Boston Dynamics Spot (Military Variants)
USA — Boston Dynamics / Various Integrators
Spot's military research program spans ISR (Intelligence, Surveillance, Reconnaissance), NBC (Nuclear, Biological, Chemical) detection, and facility inspection. While Boston Dynamics has maintained a public commitment against weaponizing Spot directly, third-party integrators and DARPA-funded research programs have tested armed configurations. The platform's advanced locomotion and expanding sensor ecosystem make it a natural dual-use candidate. Multiple NATO allies are evaluating Spot for frontline reconnaissance.
Quadruped ISR NBC Detection NATO Evaluation
03 Testing
Phantom MK-1
USA / Ukraine — Various Defense Contractors
The Phantom MK-1 represents the most aggressive humanoid combat robot concept in active testing. Designed to carry a standard military rifle, perform door-breaching operations, and navigate the kind of complex indoor environments that quadrupeds struggle with, it is being evaluated by both Pentagon-affiliated programs and Ukrainian defense innovation networks. Its humanoid form factor allows it to use human-designed environments and equipment without modification — a significant tactical advantage over purpose-built robotic platforms.
Humanoid Rifle-Capable Door Breaching Pentagon + Ukraine
04 Testing
Russia Marker UGV
Russia — Android Technica / Advanced Research Foundation
Russia's Marker is a tracked unmanned ground vehicle developed by the Advanced Research Foundation — Russia's equivalent of DARPA. Armed with anti-tank missiles, a machine gun, and RPG systems, it is designed for autonomous navigation in urban environments. The Marker uses computer vision for obstacle avoidance and target detection, and has been tested in scenarios simulating Ukrainian urban terrain. Russia has announced intentions to deploy the Marker in Ukraine, though Russia's broader war losses have slowed its industrial production pipeline.
Tracked UGV Armed Anti-Tank Urban Navigation
05 Testing
China Autonomous Combat Robots (Norinco / PLA)
China — Norinco, CASIC, PLA Ground Force
China's autonomous ground program spans multiple parallel tracks. Norinco's Sharp Claw quadruped robots mirror Ghost Robotics' design, confirmed by OSINT imagery of PLA military exercises. China's Military-Civil Fusion doctrine integrates civilian robotics advances (from companies like Unitree) directly into military programs. The PLA's stated doctrine of "Intelligentized Warfare" explicitly mandates autonomous ground systems for urban warfare, reconnaissance-strike complexes, and logistics. Chinese autonomous ground robots have been publicly demonstrated at major PLA exercises since 2022.
Quadruped Multiple Platforms Norinco PLA Integration
06 Deployed
Samsung SGR-A1
South Korea — Samsung Techwin / ROK Army
The Samsung SGR-A1 is arguably the longest-deployed autonomous lethal system in the world. Operational along the Korean Demilitarized Zone since 2010, it combines thermal imaging and pattern recognition to detect intruders at distances up to 4km. In its current configuration it can engage autonomously (subject to human authorization), mounting a 5.56mm machine gun and 40mm grenade launcher. The SGR-A1 represents the first mass-deployed system where AI targeting is a standard operational mode — a landmark in autonomous weapons deployment.
Fixed Sentry DMZ Operational Since 2010 Autonomous Mode
07 Operational
QinetiQ MAARS
USA — QinetiQ North America
The Modular Advanced Armed Robotic System (MAARS) is a battle-tested armed UGV used by US Special Operations forces and international customers. Its modular design accepts multiple weapon configurations including the M240B machine gun, M203 grenade launcher, and non-lethal systems. MAARS operates in both remote-control and semi-autonomous modes, with the ability to autonomously patrol predefined routes while alerting operators to detected threats. It has seen use in Afghanistan and is considered a benchmark for armed UGV reliability.
Tracked UGV SOCOM Modular Weapons Combat Tested
08 Testing
Rheinmetall Mission Master
Germany — Rheinmetall AG
The Mission Master is Rheinmetall's autonomous UGV platform available in multiple configurations: cargo (logistics resupply), protected (armored troop support), and armed (with a remote weapon station). Designed for NATO force integration, it uses LIDAR and sensor fusion for autonomous navigation in complex terrain. The Mission Master has been tested extensively by NATO allies including Canada and Australia. Its logistics variant addresses one of the highest-risk tasks in ground combat — resupply of forward units under fire — without requiring human exposure.
Autonomous UGV Multi-Config NATO Tested Logistics + Combat
09 Testing
Milrem THeMIS
Estonia — Milrem Robotics
Estonia's Milrem Robotics has produced one of Europe's most significant contributions to autonomous ground warfare. The THeMIS (Tracked Hybrid Modular Infantry System) is a mid-sized tracked UGV that has been integrated into NATO exercises across 15+ nations. It accepts weapons mounts including the Kongsberg Remote Weapon Station, as well as logistics and ISR payloads. The THeMIS is particularly notable for its participation in NATO's iMUGS (integrated Modular Unmanned Ground System) program — a framework for pan-European autonomous ground vehicle standards.
Tracked UGV NATO iMUGS 15+ Nations Modular
10 Testing
DARPA Squad X
USA — DARPA / Multiple Contractors
DARPA's Squad X program addresses a fundamental problem: infantry squads face adversaries with access to ISR and precision fires previously only available to large formations. Squad X gives every infantry squad autonomous unmanned systems — aerial and ground — that extend their awareness, target enemy unmanned assets, and provide precision non-lethal fires. The program specifically focuses on human-machine teaming: the robot understands the squad's intent and acts to support it, rather than requiring constant operator attention. This AI-driven collaborative autonomy is the model for near-term US ground combat.
DARPA Program Human-Machine Teaming Squad Integration ISR + Fires
11 Emerging
Figure AI (Military Potential)
USA — Figure AI Inc.
Figure AI's humanoid robot platforms represent the cutting edge of civilian humanoid robotics — and the closest civilian analog to a true humanoid soldier. Backed by Microsoft, OpenAI, Nvidia, and major defense-adjacent investors, Figure's robots use large language models for natural language command interfaces and computer vision for dexterous manipulation. While Figure maintains civilian-use positioning, its technical capabilities — bipedal mobility in human environments, tool use, environmental adaptation — are directly transferable to military logistics, EOD, and eventually combat roles.
Humanoid LLM-Controlled Dual-Use Potential OpenAI / Nvidia
12 Theoretical
Tesla Optimus (Military Applications)
USA — Tesla Inc.
Tesla Optimus is not a military system, but its potential military applications have been discussed at senior defense policy levels. With Tesla's manufacturing scale, AI training infrastructure via the Dojo supercomputer, and FSD (Full Self-Driving) vision stack adaptable to bipedal robotics, Optimus represents a platform that could theoretically be produced at industrial scale — a quantity that no adversary could match. Defense analysts note that a nation capable of mass-producing capable humanoid robots at automotive industry scale would hold an unprecedented asymmetric advantage.
Humanoid Industrial Scale Theoretical Military FSD Vision Stack

How Ground Robots Change Infantry Warfare

The introduction of AI-enabled ground robots does not simply automate existing infantry functions — it fundamentally restructures the calculus of ground combat. When one side can field units that do not fear, do not tire, and do not disobey, the human factors that have governed tactics since antiquity are suddenly in flux.

Risk Redistribution

Autonomous point men, breach teams, and scouts absorb the highest-casualty tasks. Human soldiers recede to command, fires coordination, and exploitation — dramatically changing what "acceptable losses" means in planning.

Persistent Presence

Robots do not rotate. A platoon of autonomous sentries can hold a position indefinitely without the fatigue cycles that create gaps in human-held perimeters. This transforms concepts like forward operating base defense.

Bandwidth as Ammunition

Robot swarms require communications infrastructure. Doctrines are evolving toward edge-autonomous systems that can operate on last-known orders when comms are denied — pushing AI decision-making to the platform level.

Scale and Attrition

The nation that can produce robots faster than the enemy can destroy them holds a decisive advantage. Industrial production capacity becomes a direct military metric — rewiring defense economics toward manufacturing throughput.

Human-Machine Teaming

Current US doctrine (DARPA Squad X model) places robots in a supporting role: extending human awareness, absorbing first-contact risk, and providing precision enablement. The human remains the decision-maker; the machine is a force multiplier.

Urban Combat Revolution

Door-clearing, stairwell assault, and building-to-building movement are the most lethal tasks in modern warfare. Humanoid robots that can navigate human-built environments without modification could make these operations near-zero-casualty for the fielding side.

The Uncanny Valley of Killing

The decision to build machines that look like humans — and then arm them — introduces ethical dimensions that armed quadrupeds and tracked UGVs do not. When a robot walks upright, has articulated hands, and moves with human-like motion, it triggers fundamentally different psychological and moral responses from both operators and adversaries.

International Humanitarian Law (IHL) was written for human combatants and requires the ability to distinguish combatants from civilians, assess proportionality, and show mercy. Whether an autonomous system can satisfy these requirements — and whether a humanoid form changes the answer — is the central question before the CCW (Convention on Certain Conventional Weapons) Group of Governmental Experts in Geneva.

The core tension: humanoid robots are more capable in human environments, but their human appearance may constitute a violation of the prohibition on perfidy if they could be mistaken for combatants. Nations are deploying first and seeking legal clarity second.

Key Ethical Debates

  • Can a machine exercise the "judgment of a reasonable commander" required by IHL?
  • Who bears criminal responsibility when an autonomous system commits a war crime?
  • Does the uncanny valley of humanoid robots constitute psychological perfidy?
  • Should robots be programmed to surrender, take prisoners, or assist wounded?
  • Does robot warfare lower the threshold for initiating armed conflict?
  • KAIST (South Korea) ethics controversy: AI weapons research boycott by 50+ scientists in 2018 — precedent for institutional resistance.

Ukraine as the Global Testing Ground: 2024-2026

The Russo-Ukrainian war has become the most significant real-world laboratory for autonomous systems in history. The unique combination of high-intensity peer conflict, technology-literate forces, decentralized procurement, and Western industrial backing has enabled a tempo of robotic development and deployment that peacetime programs cannot match.

Ground Robot Deployments — Ukraine Theater

  • Ukrainian THEMIS variants: Milrem THeMIS-based platforms used for casualty evacuation and forward resupply in Zaporizhzhia Oblast, reducing exposure on zero-line logistics runs.
  • Phantom MK-1 evaluations: Ukrainian defense innovation ecosystem testing humanoid breach robots in cleared urban structures in Kharkiv region, 2025.
  • Russian Marker UGV: Announced operational deployments in multiple sectors; effectiveness degraded by electronic warfare on both sides denying command links.
  • Ukrainian ground robot swarms: Ukrainian startups developed tracked bomb-delivery robots (descendant of FPV drone tactics applied to ground vehicles) — used in assaults on Russian fortifications.
  • Counter-robot doctrine emergence: Both sides developing rapid-response tactics against autonomous ground systems including terrain preparation, acoustic detection, and reactive armor adaptations.
  • Key lesson: Electronic warfare remains the primary vulnerability of all autonomous ground systems. Contested electromagnetic environments force all platforms toward greater on-board autonomy.

Military Robotics: $16B Market by 2030

$16B Market Size 2030
14.2% Annual Growth Rate
$6.4B Market Size 2024
40+ Nations Investing
$900M US Army UGV Budget FY26

The military robotics market is experiencing its fastest-ever growth phase, driven by lessons from Ukraine, US Replicator Initiative funding, and a collapse in the cost of key enabling technologies (AI inference chips, battery density, LIDAR sensors). The market bifurcates between expendable systems (single-use ground robots, low-cost) and persistent platforms (high-value autonomous systems designed for sustained operations). Both segments are growing, but the expendable segment is growing faster — reflecting the attrition-heavy nature of peer conflict.

Defeating Autonomous Ground Systems

Every capable weapons system generates a countermeasure ecosystem. Autonomous ground robots are no exception. Current counter-robot approaches span electronic, physical, and terrain-based methods.

Electromagnetic Pulse (EMP)

Directed or area EMP destroys unshielded electronics across all robot classes. Modern military robots incorporate hardening, but cost constraints mean commercial-derived systems remain vulnerable. Localized EMP devices are seeing development as counter-UGV tools.

Signal Jamming

Robots reliant on human teleoperation are neutralized by broadband RF jamming. Fully autonomous systems are immune, which is why contested EW environments are accelerating the push toward on-board AI decision-making.

GPS Spoofing

Feeding false GPS coordinates causes navigation errors in robots reliant on GNSS. Systems using visual odometry and SLAM (Simultaneous Localization and Mapping) are more resistant, but these require significant on-board compute.

Terrain Exploitation

Quadrupeds and tracked robots have real terrain limits. Mud, water obstacles, rubble fields, and certain incline angles remain challenging. Infantry doctrine is adapting to use terrain preparation as an anti-robot defense layer.

Thermal / Visual Masking

AI vision systems trained on specific signatures can be defeated by thermal masking, radar-absorbing materials, and camouflage patterns that break the object silhouettes that detection models are trained to recognize.

Kinetic Intercept

Direct fire against robots is effective but requires exposure. Ukraine has documented use of FPV drones as counter-UGV weapons — a robot-on-robot dynamic that is becoming standard doctrine for both sides.

// RELATED INTELLIGENCE

DOMAINDrone Swarm Systems CASE STUDYDARPA ACE AI Dogfight OPERATIONSExercises & Wargames DEFENSE AICounter-AI Warfare Systems NATION INTELUSA Robotics & Autonomy Programs ECONOMICSDefense Procurement Tracker