military robots

Military robots: classes, missions, and the programs funding them

Military robots are uncrewed ground machines built to take on defense tasks without a person on board, and the fielded ones mostly carry, scout, and defuse rather than fight.

By Jonathan Teplitsky · reviewed against Army, DARPA, GAO, CRS, and company primary sources · updated September 6, 2026.

Scope: what counts as a military robot

This page covers ground robots, because that is where the term is used loosely enough to cause confusion.

Uncrewed aircraft are covered as drones, and uncrewed vessels under maritime unmanned systems. Keeping the domains apart matters when reading a headline, because a "military robot" story can mean anything from a bomb-disposal machine to an armed tracked vehicle.

The useful sorting question is never "how advanced does it look". It is which mission the machine is bought for, because that decides how much autonomy it needs, how much it may cost, and how much risk anyone will accept from it.

The robot classes, and what each is really for

Ground robots sort into four practical classes, and each one is judged against a different bar.

Class	Typical job	What decides success
Man-portable / EOD	Bomb disposal, tunnel and building clearing, close reconnaissance	Reliability and simple controls under stress; the most mature class
Logistics and resupply	Moving ammunition, fuel, water, and casualties between support areas and forward units	Payload, range, and whether it can move without a driver on a real route
Legged / quadruped	Perimeter patrol, inspection, sensor carrying — see military robot dogs	Endurance and whether the legs buy anything wheels would not
Armed / combat support	Scouting, screening, and weapons payloads — see robotic combat vehicles	Cost per vehicle, survivability, and doctrine; the least mature class

Editorial classification for qualitative research only.

The order of that table is close to the order of real adoption. Bomb disposal robots have been in service for decades. Armed robotic vehicles are still trying to survive their own budget reviews.

Named Military Robots Fielded Today

PackBot and TALON are the two ground robots cited by name most often in U.S. military service. Both were built for explosive ordnance disposal (EOD) and reconnaissance rather than combat.

PackBot, now sold by Teledyne FLIR after passing through iRobot and Endeavor Robotics, has deployed in 57 countries. It has helped disarm more than 70,000 improvised explosive devices, according to Teledyne FLIR's own PackBot 510 product page. Its manipulator arm lifts up to 20 kg (44 lb), climbs stairs, and streams live video back to an operator who stays at a safe standoff distance.

TALON, built by QinetiQ North America, has logged more than 2,500 units fielded for EOD work since the mid-2000s, per QinetiQ's own TALON product page. The U.S. Army has kept TALON running well past its original fielding. QinetiQ won a $90 million contract in 2019 to maintain the existing fleet, a sustainment award rather than a new-platform buy.

Both robots predate the newer quadruped and armed ground-vehicle classes. Their long service records mark explosive ordnance disposal as the mission where ground robots have earned a lasting place in the U.S. military.

Named systems outside the EOD class have a mixed record. QinetiQ's Gladiator reached Marine Corps field testing as an armed, tele-operated tactical UGV built for scout, reconnaissance, and obstacle-breaching missions. The Marine Corps chose not to put it into production.

THeMIS, a modular tracked platform from Estonia's Milrem Robotics, has fared better. Milrem Robotics has delivered units to the Spanish Army and other NATO militaries for transport, remote-weapon, and casualty-evacuation work.

Russia's Uran-9 is the clearer cautionary tale. Senior Russian military researcher Andrei Anisimov said Syria combat testing found the remote-control link held only to 300 to 400 meters, far short of its advertised 1.8-mile range. Uran-9 also lost contact with its operators entirely for stretches as long as 90 minutes, per reporting from The National Interest.

Contested resupply is the mission with real 2026 money behind it

Autonomous resupply has become the most concrete ground-robot mission because it removes soldiers from a predictable, targetable task.

A supply convoy runs a known route on a known schedule between a brigade support area and the units it feeds. Cheap drones and long-range fires have made that route far more dangerous. Taking the driver out of the cab does not remove the risk to the cargo, but it removes the person.

The current U.S. example is TALUS, the Tactical Autonomous Logistics Utility System. In August 2026 a team led by Stratom was selected to build a TALUS prototype under the Army's Project Sustainment effort, an 18-month, roughly $15.9 million prototype award. The team pairs a militarized commercial-off-the-shelf armored cab vehicle from ND Defense with a government-validated commercial autonomy stack, and the stated goal is autonomous distribution of supplies and operational energy to dispersed units in contested environments.

The detail worth keeping from that award is structural, not technical: the vehicle is a commercial truck platform and the differentiator is the autonomy software and the modular payload architecture. That is the pattern for this mission class. The chassis is rarely the hard part, and it is rarely where the durable margin sits either.

What the RCV cancellation teaches about armed robots

The armed class carries a risk the logistics class does not: it can clear the engineering bar and still be cancelled.

The Army's Robotic Combat Vehicle program ran four vendor prototypes through competition, then stopped development in 2025 under the Army Transformation Initiative, as the Congressional Research Service program summary records and the Army's own letter to the force frames. The service later restarted the effort as a cheaper commercial-vehicle competition rather than a single premium platform.

This is not new. The Army's earlier Autonomous Navigation System effort was cancelled in 2011 even though GAO found most of its hardware and software development was already complete. Ground robotics has a longer cancellation history than most defense-tech categories, and a prototype selection is a weaker signal here than it would be elsewhere.

Autonomy is the part that is genuinely hard

Off-road autonomy, not the robot body, is the technical bottleneck for every class above.

DARPA's RACER program was built specifically for fast autonomous driving in terrain with no roads, lane markings, or reliable GPS, and its published results show how far that is from a solved problem. Communications resilience is the other half: a robot that needs a continuous link to an operator has a failure mode an adversary can create on demand.

The Army's human-machine integration work frames the goal as pairing soldiers with robotic platoons rather than replacing soldiers, which is a useful corrective when reading vendor language about autonomous formations.

Investor read-through

Military robots are an investable theme only through the companies that supply the platforms, the autonomy stacks, and the sustainment behind them.

Strong signal: a named program office, a funded prototype with a stated duration and value, and a path into a unit rotation or program of record.
Weak signal: a trade-show vehicle, an autonomy claim with no field-test disclosure, or a marketing video of a swarm with no buyer attached.
Class matters: logistics and EOD programs face engineering risk; armed programs face engineering, doctrine, and budget risk at once.
Research bridge: unmanned ground vehicle companies, robotics stocks, and the defense stocks hub.

Read this page with unmanned ground vehicles for the vehicle taxonomy and how to invest in defense tech for the prototype-versus-production distinction that decides most of these calls.

Military robot FAQs

What are military robots?

Military robots are uncrewed machines built to carry out defense tasks without a person on board, covering ground vehicles, legged robots, and explosive-ordnance-disposal machines. On this site the term is used for ground systems; uncrewed aircraft and vessels are covered separately as drones and maritime unmanned systems.

What do military robots actually do today?

Most fielded military ground robots do dull, dirty, or dangerous jobs rather than fighting. The largest real-world roles are explosive-ordnance disposal, reconnaissance into buildings and tunnels, route and site security, and carrying supplies. Armed roles remain mostly prototype work.

Are military robots autonomous?

Most are teleoperated with narrow autonomy for specific tasks such as following a route, holding a formation, or avoiding obstacles. Full autonomy in unstructured off-road terrain with degraded communications is still an open research problem, which is why programs keep a human operator in the loop.

Does the U.S. Army use armed robots?

The Army has tested armed ground robots as prototypes but does not field them at scale. Its main armed effort, the Robotic Combat Vehicle program, stopped development in 2025 under the Army Transformation Initiative before any production decision, as the Congressional Research Service summary of the program records. Logistics and reconnaissance robots have progressed further than armed ones.

What is an autonomous resupply robot?

An autonomous resupply robot is a ground vehicle that moves ammunition, fuel, water, and other supplies forward without a driver in the cab, so fewer soldiers sit in a predictable convoy. The Army funded a current example in August 2026 when a team led by Stratom was selected to build the TALUS prototype for its Project Sustainment effort.

Which companies build military robots?

The field splits between defense primes with vehicle programs, robotics specialists, and small autonomy software companies that supply the driving stack rather than the chassis. Because the same robot often carries one company's vehicle and another company's autonomy, a single program headline can be counted by more than one vendor. The unmanned ground vehicle companies page maps who does what.

How should investors read a military robot program?

Check which mission class it serves, then whether the money is a prototype agreement or a program of record. Logistics and EOD programs have a clearer path to sustained buying than armed platforms, which carry policy and doctrine risk on top of engineering risk. A demonstration video with no named program office is not an investable signal.