AI and Robotics in the U.S. Marine Corps: MARSOC’s High-Tech “Devil Dogs”

The United States Marine Corps is undergoing a technological transformation, embracing artificial intelligence (AI) and robotics to enhance its warfighting capabilities. Nowhere is this more evident than in its elite Marine Forces Special Operations Command (MARSOC), where high-tech tools are augmenting the legendary prowess of the “Devil Dogs” – a nickname Marines earned during World War I. From autonomous drones in the sky to robotic “dogs” on the ground, Marines are deploying cutting-edge systems to increase their lethality, improve situational awareness, and protect personnel. This article examines how the Marine Corps (with MARSOC at the forefront) is leveraging AI and robotics across domains, why these technologies are crucial for future conflicts, and what challenges and ethical considerations come with this new era. In a sense, the Devil Dogs now have their own “devil dogs” – AI-powered robot canines and other autonomous teammates – joining them on the battlefield.

A New Era of Warfare: Why Marines Are Turning to AI and Robots

Modern conflicts have shown the decisive impact of drones, networks, and AI-driven systems, pushing the Marine Corps to innovate or risk falling behind. Former Commandant Gen. David Berger’s Force Design 2030 modernization plan explicitly calls for accelerating the integration of unmanned systems and AI across the force. In a 2023 update to Force Design, Marine leaders introduced the concept of “Intelligent Robotics and Autonomous Systems” (IRAS) as a focal point for experimentation and acquisition. The document warns that Marines “must fight at machine speed or face defeat at machine speed,” underscoring that in tomorrow’s battles, decision cycles will be dramatically faster. This imperative comes from hard lessons of recent wars. Unmanned drones and loitering munitions played a pivotal role in the 2020 Nagorno-Karabakh conflict and in the ongoing war in Ukraine, where cheap drones and AI-driven targeting have given small units outsized punch. Peer adversaries like China and Russia are also racing ahead with armed robots and autonomous platforms, from swarming drones to robot tanks.

To prepare for these challenges, the Marine Corps is reshaping its force structure and training. It has stood up a small IRAS office led by Maj. Keenan Chirhart to bring order to the many unmanned initiatives across different Marine communities. This office is tasked with developing a unified strategy for robotics, streamlining acquisitions, and even creating new career paths. “The same way a squad leader trusts his or her Marine, they have to trust his or her machine,” said Berger, emphasizing that Marines will need to place as much confidence in robotic teammates as they do in human comrades. A major goal is to field multi-domain control systems that allow a single operator to command air, ground, and maritime unmanned vehicles through one interface. The Marine Corps Warfighting Lab has already been experimenting with such universal controllers, aiming to make managing drones and robots as intuitive as calling in a fire team support. Unifying control not only simplifies operations but reduces the number of specialists needed, which is vital for small units like those in MARSOC that deploy with compact teams.

At the heart of this push is the recognition that AI can ingest and analyze the flood of battlefield data far faster than humans. According to SOCOM’s Chief Digital and AI Officer, the goal is to harness data and machine learning “to drive split-second decision-making” for special operations forces. In other words, AI algorithms can sift through sensor feeds, identify threats or key information, and present options to operators in real time. This machine-speed processing is becoming essential as the battlefield grows more saturated with sensors and networked systems. Nowhere is this more true than in MARSOC’s missions, which often involve small teams operating in data-dense environments. Whether raiding a terrorist hideout or scouting a defended coastline, Marine special operators face a deluge of drone video, signals intelligence, and tactical reports. AI tools promise to fuse these streams and lighten the cognitive load on human teams, so they can act faster and with greater certainty.

Force Design 2030 and Integrating Intelligent Systems

The Marine Corps’ broader Force Design 2030 effort provides the blueprint for weaving AI and robotics into the force. Under this plan, traditional heavy equipment is being traded out for lighter, more high-tech capabilities better suited to dispersed operations in the Pacific and other forward environments. A key aspect is “bolder experimentation” with new autonomous systems to refine how they fit into Marine units. The Marine Corps Warfighting Laboratory (MCWL) has been charged with prototyping and testing everything from unmanned ground vehicles to AI-enabled decision aids, then feeding those lessons back to refine requirements. This rapid spiral of test and feedback is meant to overcome the notoriously slow defense acquisition cycle and keep pace with commercial tech development.

One concrete step was the creation of an experimental battalion (called IBX-30) to serve as a testbed for unmanned tactics. In late 2023, the 3rd Battalion, 4th Marines was outfitted with an array of robotic systems and participated in Exercise Apollo Shield at Twentynine Palms. In this trial, Marines controlled multiple unmanned platforms simultaneously during force-on-force maneuvers. Notably, they used a wearable controller known as the Radio Agile Integrated Device (RAID) plate to command up to four drones or unmanned vehicles at once. This modular tablet-like controller allowed a single Marine to act as a “quarterback,” coordinating a swarm of ground robots, loitering munitions, and quadcopters in concert. The ability to supervise several autonomous systems in unified fashion reduces the number of dedicated operators and increases responsiveness on the battlefield. According to Maj. Steven Atkinson, head of Robotics and Autonomy at MCWL, such advances “not only mitigate risks to Marines, but also amplify our force’s ability to respond to multifaceted operational demands rapidly and efficiently”. In Apollo Shield, the Marines even demonstrated an unmanned logistics convoy, with a self-driving convoy of robotic vehicles (including a Polaris MRZR and a Rheinmetall Mission Master UGV) ferrying supplies autonomously across miles of desert trails. Offloading routine supply runs to driverless vehicles could free up Marines for other tasks and reduce exposure to ambushes on supply lines. Taken together, these experiments are informing how future Marine units might fight with “a team of robots” organically built into their structure.

To manage this influx of autonomous tech, the Marine Corps is also considering new military occupational specialties (MOS) and training pipelines. Right now, the Corps has separate roles for things like Small Unmanned Aerial System operators and MQ-9 Reaper drone pilots, but no unified field for ground robotics or cross-domain autonomy experts. Maj. Chirhart’s IRAS office has floated the idea of a cross-domain unmanned systems operator MOS, essentially creating “robotics NCOs” who are skilled in deploying a variety of drones and bots. Such specialists would advise commanders on what unmanned tools to use for a given mission and ensure the Marines are trained to use them. “We’re not trying to replace the warfighter,” Chirhart explained. “But we know in the future, the warfighter is going to be augmented by teams of robotics and autonomous systems that operate in all domains”. Building that human expertise now – through new training programs and career tracks – is vital to actually realize the potential of the technology.

At the strategic level, the Marine Corps is forging partnerships with industry, academia, and even other services to stay on the cutting edge. The Corps is working closely with the Navy on unmanned naval systems and coordinating with the Pentagon’s Joint AI Center for enterprise-level AI tools in intelligence and logistics. They have also entered into research agreements – for example, a recent five-year collaboration with a private firm to develop AI-powered intelligence platforms for special operations. By plugging into a broader innovation ecosystem, the Marines hope to leverage commercial breakthroughs (in AI algorithms, robotics, sensors, batteries, etc.) and adapt them for military use quickly. The sense of urgency is palpable; as Maj. Chirhart noted, a procurement started today wouldn’t be fielded until 2026 at best, an eternity in tech terms. Thus, the Corps is trying to collapse development timelines by experimenting first, identifying promising prototypes early, and fast-tracking those that work.

MARSOC 2030: “Cognitive Raiders” and Tech-Augmented Teams

Within the Marine Corps, MARSOC stands out as an incubator of innovation. Marine Raiders operate in small, specialized teams often independent from larger formations, so they prize any technology that can lighten their load or expand their capabilities. In 2023, MARSOC published its own vision document called MARSOF 2030, emphasizing the need for “the Cognitive Operator” – Marines with advanced training and smart tools who can solve complex problems in austere environments. MARSOC Commander Maj. Gen. Matthew Trollinger described future Marine Raiders as “cognitive raiders”, armed not just with rifles but with data, algorithms, and networked sensors to out-think and out-maneuver the enemy. In practical terms, this means smaller teams with higher technological proficiency and gear. During a panel at the Modern Day Marine conference, MARSOC operators (speaking anonymously due to the sensitivity of their roles) underscored that they are actively seeking tools to reduce both physical and cognitive burdens on the team. For example, one critical skills operator bemoaned “one-trick pony” devices that clutter their packs – like carrying separate gadgets for mapping, laser targeting, translation, etc. – and advocated for integrated multi-function devices or software that consolidates capabilities. A communications specialist on the panel pointed to a prototype Rapid Accessory Integration Device (RAID) plate (the same concept mentioned earlier) that combines various radios and controllers into one body-worn plate. Such innovations could eliminate the need to juggle multiple handsets and control units in the field. “It could make my life 10 times easier,” she noted, by simplifying both gear and training – so that even non-communications experts on the team can manage it if needed.

Marines in MARSOC are also looking for better data fusion and analytics tools to support intelligence preparation. Currently, their all-source analysts often have to use several separate applications and databases that don’t talk to each other, requiring tedious manual correlation. One MARSOC intelligence Marine described having “five different screens up” to build the intel picture and suggested a web-based application that aggregates data in one place. This is precisely where AI can play a role – by pulling information from disparate sensors and sources into a coherent common picture. MARSOC’s Assistant Chief of Staff for Plans, Col. Ian Fletcher, highlighted artificial intelligence as an overarching solution for many of the command’s needs. The first step, he cautions, is properly defining what AI should and shouldn’t do: “We have to redefine it…as a decision support or orientation tool, not an end-all-be-all”. In other words, MARSOC views AI not as replacing human judgment, but as accelerating the “observe” and “orient” phases of the decision cycle. Fletcher noted that in high-pressure situations over the last 20 years, decisions were often made off a single input – for instance, a Predator drone feed (“strike TV”) might be the only sense a commander used to make a call on a target. AI could help incorporate many more inputs (imagery, signals, human reports) in parallel, giving a richer picture faster than a human staff could manage unaided. However, MARSOC is careful not to let AI become a crutch that erodes fundamental skills. In training, they’ve “pulled physical maps back out” to ensure Marines remain skilled in traditional navigation and visualization. The goal is a well-educated operator who can leverage AI suggestions but still build their own mental model of the battlefield.

Critically, MARSOC is aligning its tech development with both SOCOM and the broader Marine Corps, rather than building one-off solutions. “We are not developing any AI systems independently,” Fletcher emphasized, “but generating a requirement in line with the SOCOM and Marine Corps enterprises”. This means a Raider team in the field might use the same AI-enabled mapping tool as an infantry battalion on a Pacific island – ensuring compatibility and shared learning. MARSOC is taking inspiration from real-world battle labs like Ukraine, where necessity has driven rapid innovation. The command has studied how Ukrainian units have combined AI “space engines” for data fusion and tip-and-cue between sensors, achieving a level of automation in target acquisition that U.S. forces are keen to match. In meetings with Ukrainian ISR teams, MARSOC personnel discussed how AI helps speed up the find-fix-finish cycle for targets. For instance, machine vision algorithms can flag potential enemy vehicles in drone footage in seconds, or predict where a drone swarm should search next based on prior detections – tasks that would take an intel Marine much longer. MARSOC experimented with Project Maven, the DoD’s well-known AI program for analyzing full-motion video, via a Palantir interface. They also examined “homegrown” Ukrainian systems like the Delta situational awareness software, which crowdsources battlefield data and uses AI to prioritize alerts. These explorations are helping MARSOC define its requirements for AI-enabled mission support tools.

Even as they push the envelope, the Marine Raiders remain cautious about autonomy in lethal systems. Ethical and policy constraints are front-of-mind. MARSOC officials have stressed they follow all DoD directives on autonomous weapons – meaning a human is always in the loop for any use of force. In May 2024, when news broke (via industry sources) about MARSOC testing armed robot dogs, the command quickly clarified that this capability was only under evaluation, not yet fielded. The statement underscored that weapons are just one of many potential payloads for unmanned ground vehicles and that Onyx Industries’ comments about autonomous targeting reflect the company’s offerings, not an official Marine Corps program of record. This illustrates MARSOC’s careful approach: they want to leverage everybody else’s innovation (be it a startup’s AI or the Army’s robot trials) but deliberately verify it against their unique needs and ethical standards before adopting it. In sum, the Marine Raiders are aggressively pursuing new tech to become “smarter, lighter, and more lethal”, but always with an eye to maintaining the human warrior at the center, enhanced – not replaced – by intelligent machines.

---

Unmanned Aerial Systems: Eyes in the Sky and Autonomous Overwatch

In the air domain, the Marine Corps and MARSOC have rapidly expanded their use of unmanned aerial systems (UAS), ranging from hand-launched drones to large unmanned aircraft. These aerial robots are often the first foray into robotics for ground Marines, providing that invaluable “eye in the sky” for situational awareness. Every Marine infantry squad today has access to small quadcopters or fixed-wing drones for reconnaissance. Systems like the InstantEye quadcopter and the RQ-20 Puma have been fielded to platoons, allowing Marines to scout over the next hill or around an urban corner without exposing themselves to enemy fire. These mini-drones feed live video to tablets, and increasingly, AI-driven software can flag objects of interest in those feeds – for example, highlighting a possible enemy fighter hiding in shadows or pinpointing a vehicle movement in a crowded street. Such capabilities stem from initiatives like the Pentagon’s Project Maven, which delivered machine learning algorithms to automatically detect people and vehicles in full-motion video. Maven was used against ISIS and has since been integrated into many ISR platforms across the military. For Marines, this means a single operator can manage a drone patrol and let the AI alert them when something notable appears, instead of staring at hours of mundane footage. In high-tempo special ops missions, that can be a game-changer – freeing the Marine to concentrate on other tasks until “ding!” – the AI finds a target.

MARSOC units, in particular, leverage UAS heavily for both intelligence and kinetic operations. They often operate in austere environments where traditional support like manned aircraft may not be immediately available. In those cases, organic drones become lifelines. Marine Raiders have used systems like the RQ-11B Raven and RQ-21 Blackjack for mid-range surveillance, and they coordinate with Marine aviation units to get feeds from larger drones. In fact, back in 2015, a Marine Unmanned Aerial Vehicle squadron (VMU-2) deployed to support a MARSOC training exercise, flying RQ-7B Shadow drones as overwatch during a night raid scenario. The Shadows provided real-time infrared video, relayed communications, and even laser-designated targets for inbound strike aircraft. One Marine described the UAVs as a “guardian angel” overhead for the raiders – able to see the entire objective area and warn of enemy movements. Today, the Marines have gone a step further by acquiring the MQ-9A Reaper, a large hunter-killer drone long used by the Air Force. In 2023 the Marine Corps stood up its first MQ-9 squadron, and by 2025 it plans to operate a fleet of Reapers. These long-endurance drones (capable of 20+ hours flight) are loaded with sensors and can carry precision munitions. While not MARSOC assets per se, Reapers can be assigned to support Marine special operations in the field. They bring big-data capabilities: high-definition full-motion video, radar imaging, electronic eavesdropping, etc., all of which can feed into AI analytics to produce targeting intel. For example, an MQ-9 might orbit above a target compound for days, using algorithms to map patterns of life – who comes and goes, which routes are used – giving Raiders detailed intel before they ever step foot in the area.

Another emerging tool in the aerial arsenal is the loitering munition (sometimes dubbed a “kamikaze drone”). The Marines have begun fielding devices like the Switchblade 300/600, backpackable drones with an explosive warhead that can be flown directly into a target. These blur the line between missiles and robots – essentially flying smart bombs that can circle until a target is spotted. AI again plays a role: some loitering munitions incorporate image recognition to track vehicles, or can navigate to pre-programmed GPS points autonomously and search for enemy signals. MARSOC teams, operating far from artillery or air support, find loitering munitions especially handy as a self-contained precision strike asset. A Raider team inserted deep in hostile territory could launch a loitering munition from cover, have it quietly scout a compound from above, then dive onto a high-value target once identified – all within minutes and with minimal exposure. Marines recently tested integrating these munitions with their Long-Range Unmanned Surface Vessel (LRUSV) program, effectively creating drone boats that launch drone missiles. In July 2023, the Corps demonstrated an LRUSV carrying eight Israeli-made Hero-120 loitering munitions, which can accurately track and destroy targets at roughly 60 km range. The LRUSV itself is semi-autonomous: a sleek, 40-foot unmanned boat capable of extended travel, meant to move with stealth and then dispatch its aerial loitering weapons when needed. This gives Marines a potent new way to reach out and hit threats from the sea – for instance, engaging an enemy radar site on an island or a fast-attack craft menacing a naval formation. As Lt. Gen. Brian Cavanaugh explained during the demo, the LRUSV will serve primarily as an ISR and strike platform for distributed operations, aligning perfectly with the Marines’ island-hopping, littoral warfare concepts in Force Design 2030. Five LRUSV prototypes have already been delivered to the Marines by builder Metal Shark, and more are on the way. They can operate autonomously or be optionally manned, and work in tandem with a manned support boat that can house human operators when needed. In essence, the Corps is developing drone fleets that span air and sea, where a Marine back on a ship or ashore can launch a surface robot that in turn launches flying robots – all feeding target data back through an AI-enabled network.

The net effect of these aerial robotics advances is that Marines, even at small-unit level, have unprecedented situational awareness and strike options. A MARSOC team can plan a raid with a full motion video “stack” of drones overhead, use quadcopters to check every angle as they move, and call in a loitering munition or guided bomb on any threat they uncover, all within their organic capacity. This persistent overhead presence of unmanned eyes and the rapid closing of the kill-chain would have been science fiction a generation ago; today it’s becoming standard operating procedure. The challenge now is managing all that information and making sure it’s actionable. That is where AI is increasingly important – to help Marines prioritize and make sense of drone-derived intel. For example, the Marines are exploring AI tools that can auto-generate maps and 3D models from drone imagery, which operators can then use to rehearse missions via augmented reality goggles. They are also working on reliable target recognition AI that can take a feed from a surveillance drone and, without human prompting, identify enemy weapons or equipment (like distinguishing an artillery piece from a farm tractor). Each improvement shaves minutes off reaction time and could save lives in the fast, fluid engagements Marines expect in the future. As one Marine UAV commander put it, “Everyone uses UAVs because we can provide so many assets. We are guardian angels up in the sky”. With AI analytics, those guardian angels are becoming even more watchful and responsive.

Ground Robotics: From Pack Mules to Armed “Devil Dogs”

On the ground, the Marine Corps is experimenting with a variety of robotic platforms to support and protect its troops. These unmanned ground vehicles (UGVs) range from tank-like tracked machines, to wheeled cargo carriers, to four-legged “robot dogs.” For a Corps known for humping heavy loads across tough terrain, one immediate appeal of ground robots is as mechanical pack mules to haul gear, ammunition, and supplies. Earlier efforts like the Legged Squad Support System (LS3) – a headless robot resembling a mule or bull – showed promise in carrying hundreds of pounds of gear, but were ultimately set aside due to noise and complexity. However, technology has improved, and new entrants are being trialed. In late 2023, Marines with 3rd Littoral Logistics Battalion ran field tests in Hawaii with a six-wheeled UGV called the Hanwha Arion-SMET, under a foreign comparative testing program. The Arion-SMET can carry 1,200 pounds of cargo and even provide mobile electrical power for charging batteries in the field. Over two weeks, Marines put it through its paces on beaches, in jungles, and over rough trails, using modes like remote-control driving, waypoint navigation, and even a “follow me” function where the robot automatically trailed behind a leader like a pack animal. This UGV is similar to the Army’s Squad Multipurpose Equipment Transport (SMET), and reflects a joint interest in reducing the physical burden on infantry. As one Marine Systems Command leader noted, such robots “can supplement tactical vehicle operations” and move to points of need quickly without a human driver. The Marine Corps hasn’t fielded a permanent logistics UGV yet, but data from tests like this will shape requirements going forward. The vision is that future rifle squads or special ops teams might be accompanied by a small convoy of autonomous wagons or four-legged carriers, silently tagging along with extra ammo, water, and medical supplies. If firefight breaks out, the robots could bring forward reloads or even evacuate casualties. And in a pinch, these “mules” might be fitted with weapon stations to provide suppressive fire. In fact, the Army’s follow-on plans for SMET include adding modular mission payloads like jammers, sensors, or weapons – ideas the Marines are certainly watching closely.

Beyond logistics, the Marines are very interested in using UGVs to probe dangerous areas and perform high-risk missions that would otherwise endanger troops. This is where the concept of armed unmanned ground vehicles – often embodied by the agile, Quadrupedal robots nicknamed “robot dogs” – comes into play. In 2023–2024, the Marine Corps started testing these futuristic devices in earnest, in some cases making headlines around the world. One line of effort was led by the Marine Corps Warfighting Lab and Office of Naval Research, which evaluated a commercially available quadruped (the Unitree Go1 robot – a dog-sized Chinese-made robot) as a mobile weapons platform. In September 2023, at a range in Twentynine Palms, California, Marines rigged this “robotic goat” to fire an M72 light anti-tank rocket launcher as a proof-of-concept. The experiment, conducted by the Tactics and Training Exercise Control Group (TTECG) with ONR scientists, was the first known live-fire of a rocket from a legged robot by U.S. forces. Videos show the quadruped, braced by sandbags, successfully launching the 66mm rocket while a Marine controlled it from a safe distance. Afterwards, the Marine approached and reloaded the launcher on the robot’s back, demonstrating how such a system might be serviced under cover. “Instead of having a Marine handle the weapon system… we could put a remote trigger mechanism on it” allowing it all to be done remotely, explained 1st Lt. Aaron Safadi, who leads TTECG’s emerging technology section. The idea is that a Marine could stay behind cover, and send a robotic platform forward – into the kill zone or around a blind corner – armed with a disposable rocket launcher to take out a threat. Clearing bunkers, trenches, or even city streets could be made much safer by using a robot as the point man. The “robotic goat” earned its nickname from the Marines, perhaps due to the particular model’s appearance or whimsy, but effectively it’s a stand-in for any legged robotic system capable of carrying a payload. This demonstration validated the concept that even a relatively low-cost, off-the-shelf robot could deliver a knock-out punch. It opens the door to more rugged militarized versions doing the same with greater reliability.

Indeed, parallel to the ONR experiment, MARSOC itself has been evaluating more advanced quadrupedal UGVs with integrated weapon systems. According to industry reports confirmed at a Special Operations event in May 2024, MARSOC obtained at least two “robot dogs” (Ghost Robotics Vision 60 models) equipped with Onyx Industries’ SENTRY remote weapon stations. These Vision 60 robots are mid-sized, all-weather mechanical canines known for their stability and endurance – the same base model that the Air Force has tested for security patrols. What MARSOC added was a lethal payload: one robot was fitted with a 7.62x39mm rifle, the other with a 6.5mm Creedmoor rifle, both mounted on Onyx’s stabilized turret that includes an AI-driven optical system. The SENTRY system gives the robot an AI-assisted “aimbot” – a digital imaging suite that can autonomously detect, track, and lock on to potential targets (vehicles, people, or drones) within its field of view. Importantly, it cannot fire on its own; it will alert a human operator once it has a target in its sights, and the human must decide whether to engage. In essence, the robot dog roves or stands guard, and if an enemy is spotted, it paints the target and asks permission to shoot. The human controller can be remote – potentially back at a command post, since SENTRY can operate over any network and even be controlled from “anywhere in the world” via secure link. This capability could be a literal lifesaver in scenarios like cave clearance, tunnels, or dense urban blocks where the enemy could be around any corner. MARSOC has reportedly been using these armed robots in testing for tasks like “tunnel work as well as perimeter security,” per Onyx’s Eric Shell. The robots’ ability to squeeze into tight spaces where a human or even a dog might not fit – yet with a weapon at the ready – is a tremendous advantage for special operators who often must clear confined, pitch-dark networks of rooms or subterranean complexes. It’s easy to imagine a Marine Raider team sending a robot dog first into a cave used by insurgents: the Vision 60 trots in with its sensors scanning, perhaps even using thermal cameras to spot hidden tripwires or booby traps. If a hostile pops up, the robot’s AI locks on, the remote operator approves, and the threat is neutralized before the Marines ever come face-to-face. These scenarios, once limited to science fiction or video games, are becoming reality. As The War Zone quipped, the Marine Raiders may be the first in the U.S. military to field “rifle-wielding robot dogs” in an operational context.

For the Marine Corps at large – proud “Devil Dogs” since 1918 – there is a certain poetic justice in now having actual robotic dogs join the ranks. The Corps even uses a bulldog mascot (Chesty) to symbolize its tenacity. With the advent of these AI-driven canines, the Devil Dogs truly have their own devilish dogs on the battlefield. The nickname isn’t just metaphor anymore. Marines have already started informally calling some of these machines “Teufelhunden” in homage to the famous moniker, underscoring how quickly robots can become part of the service’s culture and lore. The pairing of Marines and tough, reliable dogs has a long history (think of Marine K9 handlers and their heroic bomb-sniffing dogs). Now the Corps is extending that camaraderie to mechanical dogs that feel no fear or fatigue. One Marine dog handler, upon seeing the Ghost Robotics quadruped in action, reportedly joked, “The only thing it won’t do is lick your face – otherwise it’s a darn good dog.” These robots can patrol perimeters endlessly, brave minefields, or stand guard in freezing cold or searing heat where it would be cruel to send a real dog. And unlike a flesh-and-blood canine, a robot dog armed with a heavy weapon can lay down suppressive fire or knock out a light armored vehicle, all while keeping Marines out of harm’s way.

From a capabilities standpoint, the armed UGVs bring several clear advantages:

• Hazardous Duty: They can be used for jobs like clearing mined areas, entering tunnels or buildings first, and absorbing potential ambushes or IED blasts, removing Marines from the immediate danger.
• Persistent Security: As sentries at a forward operating base or expeditionary site, they don’t get tired. The Air Force has already deployed unarmed versions at bases like Nellis AFB for security patrols, and the Marines see similar value for perimeter security in remote outposts.
• Mobility: Quadrupeds can traverse rough terrain – rubble, stairs, narrow alleys – better than wheeled robots. This agility is crucial in urban warfare and complex terrain.
• Integrated Sensors: Even without a weapon, these robots serve as sensor hubs, carrying cameras, thermal imagers, and communications relays. An unarmed configuration can quietly scout and map areas, essentially acting as a forward reconnaissance scout for the unit.
• Reduced Casualties: By taking point or handling a fight independently, they reduce the risk to human Marines. As Shell from Onyx noted, you keep the operator “out of danger” – potentially even on the other side of the world – while the robot confronts the threat.

Of course, there are limitations and concerns as well. These systems currently are under evaluation and not yet standard issue. They rely on robust communications; jamming or loss of link could neutralize their effectiveness. They are also expensive (the Ghost Robotics dog runs in the hundreds of thousands of dollars with its payload) and delicate relative to a warfighter – though hardened for military use, a well-placed shot can disable a robot dog, whereas a human Marine might apply a tourniquet and fight on. Moreover, the Marine Corps is being very deliberate about keeping humans in control of lethal force, abiding by DoD policy that requires meaningful human oversight over weapons decisions. The ethical debate is ongoing: autonomous targeting is here, but how far should it go? As one Ars Technica piece noted, the fielding of armed robot dogs seems “likely inevitable at this point,” and it raises tough questions about “how long a human will stay in the loop” as AI targeting improves. For now, the Marines are erring on the side of caution – using the AI to identify targets, but reserving kill decisions for a Marine. In testing, they are also examining non-lethal uses of the same tech, like mounting only sensors or electronic warfare modules on the quadrupeds, which could sniff out enemy signals or jam communications without any kinetic action.

Meanwhile, not all ground robots are so futuristic – some are quite familiar. Marine Corps explosive ordnance disposal (EOD) teams have used tracked bomb-disposal robots (such as the Mark II Talon and newer models) for decades. Those robots, essentially remote-controlled mini-tanks with an arm, have saved countless lives by dealing with IEDs and unexploded ordnance in Iraq and Afghanistan. What’s changing now is that such robots are getting smarter and more autonomous. New EOD robots can use AI to autonomously navigate to a bomb and even attempt certain neutralization procedures under supervision. There is cross-pollination between EOD and infantry robotics; for instance, a small tracked robot initially designed to inspect explosives might also be used by infantry to scout a building (some are small enough to toss through a window, like the Throwbot or similar devices). As autonomy improves, the Marines envision teams of ground robots – some tracked, some wheeled, some legged – working together. One can imagine a breaching scenario where a small rolling robot blows open a door, a robot dog dashes in to secure the room and identify friendlies vs hostiles with AI facial recognition, and a heavier armed UGV rumbles in behind to provide fire support. While such coordinated autonomy is still in development, the pieces are being assembled through the various experiments described.

In short, the Marine Corps is moving steadily toward a future where every infantry platoon or special ops team has a set of robotic wingmen. These “Devil Dogs” of steel and silicon will carry ammo, scout ahead, stand guard, and if necessary, take the first shots in a fight. It’s a profound change – effectively mechanizing and digitizing the small-unit fight without adding humans or vehicles, but by adding smart machines. Marines have always prided themselves on doing more with less; with AI and robots, a 14-Marine squad might do what used to require 40 Marines and a convoy of vehicles. As these tools come online, MARSOC and other early adopters are refining tactics for their use. They report that one of the biggest challenges is simply trust and familiarity: Marines must learn to trust a robot to cover their flank or clear a hallway. Training is addressing this by embedding robots in exercises as if they were new team members, letting the Marines work out communication, timing, and confidence in what the machine can do. According to those involved, the young generation of Marines takes to it quickly – after all, they’ve grown up with technology and even the concept of drones and bots in gaming. As one Marine quipped during a demo, “It’s like having a Pokémon with a machine gun – once you get over the cool factor, you just want to know what else it can do for you.”

Maritime Autonomy: Unmanned Boats and Underwater Systems

Given the Marine Corps’ naval roots, it’s natural that they are also exploring robotics in the maritime realm. The aforementioned Long-Range Unmanned Surface Vessel (LRUSV) is a prime example of Marines partnering with Navy and industry to create drone boats that support amphibious operations. These vessels essentially give the Marines their own unmanned naval capability for the first time. They are modular: in addition to carrying loitering munitions, LRUSVs can potentially launch small UAVs, act as decoys, perform reconnaissance, or serve as communications relays for forces ashore. In recent tests off the U.S. East Coast, multiple LRUSVs operated together, suggesting the Marines may employ them in packs to saturate an area with sensors or munitions. The Corps is also interested in unmanned underwater vehicles (UUVs) for beach and littoral recon. Before a Marine Raider team conducts an amphibious landing on a hostile shore, they might deploy a UUV (like the small REMUS drones the Navy uses for mine-hunting) to survey the beach for obstacles and mines. Such UUVs use side-scan sonar and AI-based image classification to pick out mines or defenses on the seabed, marking safe lanes for the Marines. In the Pacific, where vast ocean areas and island chains dominate, the Marines (now forming Littoral Regiments) foresee using uncrewed surface and sub-surface drones to extend their eyes. Autonomous small boats could quietly loiter off an enemy-held island, snooping on radio emissions or movements, feeding intelligence back to the Marines over the horizon. The Navy is actively developing networks of such systems, and the Marines are plugged into those efforts so that when the time comes to seize a strait or chokepoint, an orchestrated dance of unmanned systems can set the stage.

For example, in an exercise, a Marine Littoral Regiment might dispatch a wave of unmanned kayaks or submersibles to creep into an enemy harbor at night, mapping targets and perhaps emplacing sensors. At dawn, an LRUSV swarm might rush in from standoff range to bombard coastal defenses with loitering drones, while overhead a Navy MQ-4C Triton (a large surveillance UAV) provides broad area monitoring. All this could happen before the first Marine in a landing craft gets near the beach. It’s a level of preparation and stand-off engagement that fundamentally changes amphibious assault tactics – making them more akin to an orchestrated, multi-domain raid with robots in the vanguard.

MARSOC, with its maritime mission sets (direct action raids, special reconnaissance, foreign internal defense in littorals), is keenly interested in these developments. Marine Raiders train extensively in combat diving and small boat operations; adding unmanned systems to their toolkit multiplies their effectiveness. A Raider team operating off a coastal village might launch a drone boat with an ISR payload to surveil the shoreline while they insert from a different angle. Or use a UUV to covertly deliver supplies or munitions to a rendezvous point (there have even been concepts of UUVs as underwater cargo shuttles). The technology for fully autonomous long-distance UUVs is still maturing, but it’s a space to watch. The Ukraine conflict recently highlighted offensive UUVs – essentially robotic kamikaze torpedoes used against ships. The Marines could face similar threats or employ their own one day as part of sea-denial operations.

All told, air, land, and sea robotics are converging to give the Marines a ubiquitous presence across the battlespace. AI serves as the glue enabling one Marine to effectively control multiple unmanned systems and process the flood of sensor data they produce. The Corps is actively working on networks and data architecture (for example, the “observability” of its networks and using AI for IT operations) to ensure these unmanned systems can reliably communicate even in contested environments. There’s a recognition that data is a strategic resource, and that having secure, interoperable networks (meeting the Pentagon’s VAULTIS principles – visible, accessible, understandable, etc.) is critical for AI and robotics to reach their potential.

Challenges and Ethical Considerations

For all the excitement around AI and robotics, the Marine Corps remains clear-eyed about the challenges. One major concern is reliability and trust. Complex autonomous systems must work as intended not just in lab settings but in mud, sand, and under fire. Marines need to trust that a robot dog won’t accidentally discharge a weapon or that an AI won’t misidentify a friendly as a foe. Rigorous testing and clear rules of engagement are therefore essential. MARSOC’s careful statement about the robot dogs – emphasizing they are not fielded yet and following all DoD policy – highlights that any introduction of semi-autonomous lethal systems will be deliberate. The Corps also knows adversaries will attempt to counter or exploit these technologies. Jamming, hacking, or even deceiving AI (through camouflage or tactics like hiding among civilians) are all concerns. Cybersecurity for robotics is critical; as Col. Fletcher noted, deploying AI in classified, secure environments is a challenge, and “AI is only as smart as the data that’s coming into it”. If an adversary feeds false data or the training data is limited, the AI’s performance can degrade. To mitigate this, the Marines are working on robust control links (potentially using encrypted mesh networks or satellite communications) and redundancy (so if one drone is lost, another can pick up the mission).

Ethically, the Corps is navigating how to use AI in decisions without losing the human moral compass. There is a line they don’t want to cross – such as having an AI outright plan missions or “write operational orders” on its own, which Fletcher explicitly said they want to avoid. The Marines see AI as an assistant, not a commander. This philosophy is in keeping with U.S. military AI ethics guidelines that stress responsibility, equitability, traceability, reliability, and governability in AI systems. Practically, this means any AI suggestions (say, target recommendations or route plans) must be transparent and explainable to the Marines using them, and ultimately a human makes the call. It also means retaining manual overrides on all autonomous platforms – a Marine can always take direct control or shut it down if something seems wrong.

There’s also the question of training and doctrine: incorporating robots changes unit tactics. Marine rifle squads might evolve to include a “tech corporal” who manages the drones or UGVs during a fight. Fire and maneuver could take on new dimensions when a pair of unmanned systems is part of the maneuver element. The Corps is updating its field manuals and training simulations to account for these possibilities, ensuring that units practice with robotics as naturally as they do with machine guns or mortars. Large exercises (like the Marine Air-Ground Task Force Warfighting Exercise mentioned in the Combat Center news) are starting to insert robotic surrogates to test how units react – and interestingly, to train Marines on how to defeat enemy robots as well. Future Marines may have to face swarms of hostile drones or unmanned vehicles themselves, so developing counter-robot tactics (using electronic warfare, directed energy weapons, or old-fashioned marksmanship) is equally important.

In terms of logistics and maintenance, high-tech gear can strain resources. The Marines are famous for their austere, expeditionary mindset, but an autonomous fleet requires spare parts, software updates, and possibly contractors or specially trained technicians. The Corps is working to make these systems as rugged and user-friendly as possible. Some robots will inevitably break down in the field; the goal is to have Marines able to fix common issues quickly (perhaps by swapping modular components). The simpler the interface, the better – one can imagine future Marines using voice commands or intuitive tablet apps to direct their machines, rather than fiddling with a keyboard. Achieving interoperability between different systems is also a hurdle: the Corps wants that universal controller to truly work across various drones and robots, which means convincing vendors to adopt common standards. The Marine Innovation Unit is actively engaged in liaising with tech companies to address such integration challenges, drawing on reservists with tech industry backgrounds to assist.

Lastly, there’s the cultural aspect. Marines have a proud tradition of grit and personal honor in combat. Some worry that relying on robots might dull those qualities or create over-reliance on technology. The leadership so far has struck a balance: they frame robotics as simply new weapons and tools. Just as the introduction of tanks or helicopters didn’t make Marines less Marine, the introduction of AI and robots is cast as an evolution that honors the Corps’ ethos of adaptability and fierceness. In fact, being among the first to master these tools is seen as a new source of pride. In motivational speeches, leaders point out that Marines have always been known as “first to fight” – now they aim to be first to innovate as well, in order to fight and win. The “Devil Dog” spirit, if anything, extends to attacking this new tech with the same aggressiveness as any obstacle. One can imagine future lore about famous Marine-robot duos, or legendary exploits of a squad that held off an enemy battalion with a handful of unmanned systems at their side.

Conclusion

The U.S. Marine Corps, and especially its MARSOC Raiders, are entering a new age where AI and robotics augment every facet of operations. Drones buzz overhead as perpetual scouts, driverless vehicles haul loads and secure the flanks, and robotic dogs armed with cameras (and sometimes guns) walk alongside their human counterparts into battle. The marriage of Marine ethos with Silicon Valley innovation is producing a formidable hybrid: tech-savvy warfighters equipped with autonomous brethren. The benefits – greater lethality, improved protection, faster decisions, and logistical agility – are driving this transformation, even as challenges of trust, ethics, and practicality are carefully managed.

In the span of just a few years, the image of a Marine patrol could shift from a purely human lineup to a human-machine team, with Marines and their robot teammates moving as one. A Raider team infiltrating a terrorist compound might consist of a half-dozen Marines plus a squad of robotic support: a quadcopter providing overwatch, a “devil dog” quad-UGV breaching the door, and an autonomous mule bringing up extra ammo. In essence, the Corps is multiplying its force without significantly increasing manpower – the extra “Marines” are machines guided by AI and loyal to the mission. This melding of man and machine on the battlefield is truly the next chapter in Marine Corps history.

As these technologies mature, we can expect to see them fielded more broadly across the infantry and reconnaissance communities, fulfilling Force Design 2030’s vision of a lighter, faster, more lethal force. The lessons MARSOC learns today with armed robot dogs or experimental AI tools will inform how the entire USMC fights tomorrow. And while war will always be a chaotic, human endeavor at its core, the inclusion of AI and robotics promises to tilt the odds in favor of those who master them. In keeping with their storied nickname, the Marines are ensuring that “Devil Dogs” will dominate not just in steadfast courage, but in technological prowess – unleashing literal devil-dogs of their own making to overwhelm the enemy.

When future adversaries face the Marine Corps, they won’t just be up against tough, determined Leathernecks; they’ll also have to contend with an array of Marines’ robotic auxiliaries – tireless, fearless, and powered by cutting-edge AI. This fusion of warrior and machine is poised to make the Marine Corps more adaptive and deadly than ever. The Devil Dogs of Belleau Wood would surely be astonished to see robotic dogs in the ranks, but they would recognize the bold, relentless spirit driving this change. In the end, it’s that Marine fighting spirit – enhanced by advanced tech – that will continue to carry the day in battles to come.

---

References

1. Parken, Oliver, and Howard Altman. “Rifle-Armed Robot Dogs Now Being Tested By Marine Special Operators (Updated).” The War Zone (TheDrive), 7 May 2024.
2. Trevithick, Joseph. “Marines Test Fire Robot Dog Armed With Rocket Launcher.” The War Zone (TheDrive), 18 Oct. 2023.
3. Seck, Hope Hodge. “Marines Overhaul Approach to Smart Robots, Eye New Military Job.” Marine Corps Times, 31 Aug. 2023.
4. Carberry, Sean. “Marine Special Operators Seeking New Tech, AI for Future Missions.” National Defense Magazine, 29 Sept. 2023.
5. Edwards, Benj. “Robot Dogs Armed with AI-Aimed Rifles Undergo US Marines Special Ops Evaluation.” Ars Technica, 8 May 2024.
6. Saballa, Joe. “US Marines Test Robot Dogs With AI-Enabled Rifles.” The Defense Post, 9 May 2024.
7. Bray, Cpl. Andrew. “Marines Test Emerging Technologies at The Combat Center.” Marine Corps Air Ground Combat Center, 19 Oct. 2023.
8. Marine Corps Air Ground Combat Center (MCAGCC). “The Combat Center Hosts Unmanned Force-on-Force Exercise Apollo Shield for IBX30.” MCAGCC News, 21 Oct. 2023.
9. Ong, Peter. “USMC Adds Loitering Munitions to its LRUSV Naval Drones.” Naval News, 18 July 2023.
10. Wikipedia. “Devil Dog.” Wikipedia, Wikimedia Foundation, last edited 7 Feb. 2025.