Military robot dogs: uses, limits, and investor signals

Military robot dogs are legged unmanned ground vehicles built to move sensors, radios, or payloads through places where wheels and tracks can struggle.

What military robot dogs are

Military robot dogs are quadruped UGVs that use four legs instead of wheels or tracks.

The U.S. Air Force describes quadruped unmanned ground vehicles as semi-autonomous robots intended to reduce human exposure to danger and improve base security. The U.S. Army describes robotic dogs as teleoperated ground ISR robots used ahead of dismounted formations.

First mentions matter here: Ghost Robotics Vision 60 is described by its maker as a quadrupedal unmanned ground vehicle for commercial and defense needs. That does not make every robot dog a defense winner, but it shows how vendors frame the category.

Why legs matter

Legged mobility matters because military terrain is often uneven, cluttered, or built for people rather than vehicles.

Speed, Payload, and Endurance by Platform

Manufacturer spec sheets put today's military-relevant quadrupeds at very different speeds, and the numbers explain why "how fast can a robot dog run" does not have one answer.

The Spot spec sheet lists 1.6 m/s as a walking speed rather than a sprint, which is slower than a person jogging. Unitree's Go2 product page notes its top speed was measured in lab conditions, a caveat worth reading literally, since field speed on broken ground runs lower. Ghost Robotics' Vision 60 page separates a 0.9 m/s standard walk from that 2.5 m/s sprint figure. A robot patrolling a perimeter for hours spends most of its runtime at the slower number.

Military uses are broader than weapon videos

Military robot dogs are often more useful as sensor carriers than as dramatic weapons platforms.

The common missions are perimeter patrol, reconnaissance, hazardous-site inspection, communications relay, and payload carriage. The information-gain point is that armed demos can distract from the real procurement question: whether the robot improves a repeatable mission enough to justify training, maintenance, and integration.

The Marine Corps has tested quadruped and heavier robotic platforms in force-on-force exercises, which is useful because field exercises expose operating limits that polished demos can hide.

What limits adoption

Military robot dogs face adoption limits around endurance, payload, communications, noise, repairability, and trust.

A military robot dog can be hacked, jammed, or spoofed in principle because any networked or remotely operated ground robot depends on wireless links and onboard software. Boston Dynamics, Unitree, and Ghost Robotics do not publicly disclose electronic-warfare hardening or cybersecurity measures. No publicly documented incident of a military robot dog being hacked, jammed, or spoofed exists as of this writing. Physical conditions like rain, dust, urban clutter, and long-duration patrols create separate operating limits. Operators also need established procedures for battery charging, transport, mission planning, data handling, and retrieval if the robot fails.

Boston Dynamics, Unitree, and Ghost Robotics do not publish a maintenance or repair cost breakdown. Evidence for recurring revenue instead requires tracking spare parts sales, service or maintenance contracts, and software subscriptions reported in regulatory filings or investor materials. None of the three currently files as an independent, standalone U.S.-listed company with its own disclosed robot-dog segment financials, so readers cannot look up this recurring-revenue detail for any of them directly.

The legged-robot research base is not a US-only asset

Quadruped mobility did not appear inside a single national program. It grew out of decades of published, openly cited research, and that history matters to an investor because it explains why more than one country's industry can build a competitive robot dog today.

The clearest example is Boston Dynamics' early quadruped work: BigDog and its follow-on LS3 pack-mule prototype were developed under DARPA funding in the 2000s and early 2010s, and the underlying research on dynamic legged balance was presented at open robotics conferences rather than kept proprietary. That publication norm is standard for DARPA-funded university and lab research, and it means the core control techniques for four-legged balance and gait were part of the global academic record well before any single company turned them into a shipping product. A funding source and a permanent competitive moat are not the same thing once the resulting research is public.

Since then, the commercial center of gravity for low-cost quadrupeds has shifted toward Chinese manufacturers, and Unitree is the clearest public example: it sells quadruped robots at price points well below early Western platforms, backed by a domestic supply chain for the actuators, motors, and batteries that make up most of a legged robot's bill of materials. That cost advantage is a manufacturing and industrial-policy story as much as a research story — a government that treats robotics components as a strategic manufacturing sector can under-price a rival whose components come from a thinner, more expensive supply base, independent of who published the original control research.

That research trail runs through one program by name. Unitree's Go-series design has been tied specifically to the Mini Cheetah, a small quadruped built by the Massachusetts Institute of Technology (MIT). MIT's Biomimetic Robotics Lab introduced it in 2019. Army Research Laboratory (ARL) funding backed that work through the Robotics Collaborative Technology Alliance (RCTA), a multi-year university and industry robotics consortium. Ben Katz, an engineer who worked on the Mini Cheetah at MIT, has said publicly that Unitree's Go series matches the Mini Cheetah's dimensions almost to the millimeter. That is one named engineer's design-similarity claim. It is not a patent ruling, a lawsuit, or a government finding of design theft, and none of those exist publicly as of this writing. Army-funded robotics research is normally published rather than kept exclusive, and that norm applies to this claim as well. A resemblance this close is consistent with close study of published results. It is equally consistent with two engineering teams reaching the same physical solution once battery, motor, and actuator choices converge on the same components.

New behaviors get added to a fielded robot dog mostly through software, not a new chassis, which is a second reason this field is hard to keep exclusive. Balance recovery, obstacle stepping, and load-carrying gaits are increasingly trained with reinforcement learning in simulation and then transferred to the physical robot, a technique published widely enough that any team with enough compute and a physical test unit can iterate on it. A cheap, widely available commercial quadruped is also a cheap platform for outside researchers and rival firms to study and improve on once it ships, which is a different and faster diffusion path than a classified military program would ever allow.

Armed quadruped demonstrations have also surfaced on both sides, and neither is evidence of a fielded program. US-based vendors and independent gunsmiths have shown weapon-mounted robot dogs at trade events, drawing public criticism and no confirmed military purchase order; Chinese state media has separately shown PLA-linked quadrupeds carrying rifle-style mounts in exercise footage. In both cases the correct read is the same one this page applies to every robot-dog headline: a demo proves the mechanical mount works, not that any military has adopted an armed quadruped as a program of record, and no major military has publicly confirmed doing so.

For a research angle, the practical takeaway is to separate three different claims that get collapsed into one narrative: who funded the original mobility research (often public, often US-funded, and openly published), who can manufacture the resulting hardware cheapest today (increasingly China, on a component cost-base advantage), and who has actually fielded an armed or mission-configured version at scale (no one, publicly, as of this writing). A company's marketing will usually only tell you which claim it wants you to hear.

Robot Dogs Around the World

China, India, Russia, and Ukraine have each shown or used quadruped robots for military or parade purposes, and the four cases sit at very different points between demonstration and deployment.

• China: The People's Liberation Army (PLA) displayed armed quadrupeds nicknamed "robot wolves" at a September 3, 2025 parade in Beijing marking 80 years since Japan's defeat in World War II. State media described the machines as built for frontline reconnaissance, supply delivery, and precision strikes, with some carrying mounted rifles, according to CBS News' coverage of the parade. Chinese state media has since shown the same robot wolves inside PLA training exercises.
• India: The Indian Army has fielded a robot dog called MULE, short for Multi-Utility Legged Equipment, built by New Delhi-based company Aero Arc. The Army ordered 100 units and received full delivery by mid-2024, with one nicknamed "Sanjay" appearing at Republic Day parades on New Delhi's Kartavya Path, according to WION's reporting. Aero Arc markets MULE for perimeter security and chemical, biological, radiological, and nuclear (CBRN) reconnaissance.
• Russia: Russia displayed a robot dog armed with an RPG-26 anti-tank rocket launcher at the Army-2022 forum, built on a platform its maker called M-81. The War Zone's reporting found the robot's frame closely matched Unitree's consumer-grade Go1 model, raising doubts it was a Russian-built military system rather than a rebadged commercial robot. No confirmed order or fielded unit has followed the 2022 demonstration.
• Ukraine: Ukraine has taken robot dogs furthest into actual combat use. British firm Brit Alliance has supplied more than 30 of its BAD2 robot dogs to Ukrainian forces. Ukraine's Khortytsia operational-strategic group has used them for reconnaissance near Toretsk, close to the destroyed city of Bakhmut, according to Forbes defense reporter David Hambling. Each BAD2 costs about $9,000 and runs at up to 9 mph. Hambling has also reported that some Ukrainian-deployed units are rebadged Unitree Go2 Pro robots rather than a Brit Alliance original design.

The same read applies across all four countries: a parade appearance or trade-show demo is not the same signal as a unit carrying the robot into the field. Ukraine's use is the only one of the four that has clearly moved past the demonstration stage.

Who Makes Military Robot Dogs

Three companies build most of the named military and government robot-dog platforms: Boston Dynamics, Unitree, and Ghost Robotics.

Boston Dynamics built the earliest military-funded quadrupeds. BigDog, its first four-legged prototype, led to the DARPA-funded Legged Squad Support System (LS3). The program nicknamed it AlphaDog, a 400-pound pack-mule robot the Marine Corps tested in the field. DARPA put the program into storage in 2015. Boston Dynamics' current commercial platform, Spot, is the model both the Air Force and the New York Police Department have tested. Boston Dynamics is privately held. Hyundai Motor Group agreed in 2026 to buy SoftBank's remaining stake for about $325 million, making Boston Dynamics a wholly owned Hyundai subsidiary. It carries no public stock ticker as of this writing.

Unitree, based in Hangzhou, China, undercuts Western quadruped prices with its Go-series lineup, built on a lower-cost domestic supply chain for motors, actuators, and batteries. Its consumer-grade Go1 model is the same design that Russian military displays have used as a base, including the M-81 robot dog Russia showed at the Army-2022 forum.

Ghost Robotics, based in Philadelphia, builds the Vision 60 that the Air Force has tested for base security. The same platform became the first robot dog publicly shown carrying an actual firearm. Nevada-based SWORD International mounted a 6.5mm Creedmoor rifle on one, called it the Special Purpose Unmanned Rifle (SPUR), and unveiled it at the AUSA trade show in 2021. A separate category of viral robot dog comes from consumer manufacturers rather than military programs. Ohio-based Throwflame sells the Thermonator, a flamethrower-equipped robot dog available to the public for $9,420. In 2024, South Korean defense electronics firm LIG Nex1 bought a 60% controlling stake in Ghost Robotics for $240 million.

None of the three companies discloses a public breakdown of military versus civilian robot-dog sales, so unit counts by buyer type are not available.

What Military Robot Dogs Cost

Boston Dynamics launched Spot's commercial price at $74,500 in 2020, covering the robot, battery, charger, and tablet controller, according to reporting at the time from IEEE Spectrum. Boston Dynamics no longer publishes a flat list price. Configured industrial units with add-ons such as a robotic arm or thermal sensor have since been reported in the $150,000 to $375,000 range.

Unitree prices its Go2 line far lower, from about $1,600 for the base Air model to several thousand dollars for EDU configurations aimed at universities and developers. That range comes from Unitree's own store listing. Neither figure is a military contract price, and none of the U.S. services that have tested robot dogs has published a per-unit or program-level procurement cost.

Ukraine's British-supplied BAD2 units are the one figure with battlefield context attached: Brit Alliance prices each robot dog at about $9,000, and Ukrainian forces have received more than 30 of them. That number describes wartime aid rather than a competitive U.S. Department of Defense acquisition program. It should not be read as a benchmark for what a U.S.-fielded robot dog would cost.

Investor read-through

No U.S. military branch has publicly confirmed a program-of-record purchase of robot dogs, as current Air Force and Marine Corps activities remain evaluation trials instead of funded production contracts. Robot dogs therefore remain one niche inside the wider UGV and defense-robotics stack. To track any future transition into formal procurement, check Department of Defense (DoD) contract announcements for confirmed production awards.

• Strong signal: official exercises, base-security use, payload integration, and named defense customers.
• Weak signal: viral videos without procurement, unclear autonomy claims, or weaponized demos without authority context.
• Research bridge: compare unmanned ground vehicle, ground drone, and robotics stocks coverage.