# Are mechanical walking vehicles useful in combat? [duplicate]

This is a highly contested issue on this site, and I have seen many different opinions on this topic. Are mechanical legged vehicles (or mechs) useful in combat in a near future scenario? If so where?

• Legged Vehicles are all terrain vehicles. They can walk over rocks, climb mountains, jump up buildings, jump over gaps, etc. See worldbuilding.stackexchange.com/questions/112144/… – cowlinator Feb 11 '20 at 3:41
• I'd consider a "legged vehicle" and a "mech" different things, personally. Although I can see some overlap, at the very least a mech is a subcategory of legged vehicles. A legged vehicle I'd assume is something like the AT-ATs in Star Wars. Although they are quite big, I guess you could have them smaller, too. While a mech is a humanoid looking on two legs with (usually) two arms and is definitely very big. – VLAZ Feb 11 '20 at 6:37
• VLAZ, that is not how I interpret mech and legged vehicle, but I can see how your point of view makes sense. probably good to clear up the difference. – meaninglessname Feb 11 '20 at 8:27
• You can check Boston Dynamics' "Spot", a four legged robot that could be used as a load carrier in combat situation (where wheels could easily be a problem). – Asoub Feb 11 '20 at 12:06
• Although the Metal Gear series is all about walking tanks, I believe in MGS3 they actually mention how totally impractical the idea really is. – pboss3010 Feb 11 '20 at 12:34

There's an interesting case for 'legged' vehicles that have wheels and treads rather than feet. DARPA's Ground Vehicle-X Technology program makes an interesting case for agile, fast, and lightweight ground vehicles. They make the case that our current approach of survivability through heavy armor has some problems. The first is that armor makes vehicles heavy and difficult to transport into the battlefield. If vehicles could be made light enough to load onto planes or helicopters, then they could be deployed into the battlefield much faster, which is huge. Vehicles are starting to become so heavy that they can no longer go offroad, which makes them vulnerable to IEDs. They also make the case that weapons are improving faster than armor is, so the advantage of armor is becoming less.

They claim the solution to this is to achieve survivability through mobility rather than more armor. A vehicle doesn't need much armor if it can avoid being hit, avoids detection, and avoids engagement. Extreme mobility enables the vehicle to access terrain that other vehicles cannot access making it possible to avoid detection and engagement. Putting the wheels on legs so that they can adapt to terrain is a great way to attain high mobility. They make the case that it's not unreasonable for a vehicle to duck or jump to avoid getting hit. This would have to be accomplished by putting the body on some form of legs so that we may move it around. In addition, it has even been suggested that jumping may be advantageous in urban combat, to jump over road blocks.

• This actually reminds me of a more accurate version of the Tachikomas of Ghost in the Shell, which use what are effectively wheeled legs for a highly mobile platform. – Adam Reynolds Feb 11 '20 at 7:24
• Ducking/ dodging seems very unplausible to me. Just consider the speed a typical anti tank missile moves at. Note that these are also targetting so they can adjust their course at least somewhat. Dodging that would require an acceleration that might be mechanically possible but any humans inside the vehicle won't survive it. – quarague Feb 11 '20 at 12:35
• This reminds me of Batman's Tumbler. youtube.com/watch?v=jZicrp_p4BE I'll take 2, please. – computercarguy Feb 11 '20 at 18:26
• "They also make the case that weapons are improving faster than armor is, so the advantage of armor is becoming less." History requires that to be taken with a hefty helping of salt. They've been saying this for decades about everything from the man-portable anti-tank missile to the attack helicopter to drones and armor has responded. "Speed is armor" has been tried from the battlecruiser to the Leopard 1 to the Mobile Gun System. These are successful as specialist, support, and scout vehicles, but do not replace the staying power of armor. – Schwern Feb 12 '20 at 3:02
• "weapons are improving faster than armor" is no longer true. In the 16th-20th century this was definitely the case, but over the past few decades, modern material sciences have been producing alloys, ceramics, synthetic fibers, carbon nanostructures, and a whole plethora of various countermeasures that have been making armor technology advance much faster than weapon tech. This is a big part of why modern tank cannons keep having to be made proportionally bigger. – Nosajimiki Feb 12 '20 at 20:07

Almost certainly not. The problem that cannot be countered is the inherent mechanical inefficiency of using legs. The amount of complex engineering that would have to go into this process is not worth the benefits. From a physical standpoint, it is inherently inefficient. There are too many moving parts under a dynamic load, with a center of mass that is far too high.

Consider what the act of walking takes in comparison to a more conventional ground vehicle. Each time you lift the leg, you are wasting energy to fight gravity in order to physically lift that leg off the ground. You then have the issue of ground pressure when you put that foot down and shift weight on top of it. Wheels or especially treads will have a much lower ground pressure because the load is always balanced.

Put it this way, would you ever challenge a bicycle to a race on foot? They are inherently physically inefficient. Technology won't change this. Whatever technology that makes them possible would also make conventional ground and naval vehicles more effective. Just like the fact that even an extremely fast runner would be even faster on a bike.

You might counter that there are places you can't take a bicycle, like stairs, but those aren't places where you want to take heavy mechanized vehicles in the first place. No one drives tanks into truly mountainous terrain or dense forests, as even if it is physically possible to move them they would be too vulnerable for it to be worth it. If you take ground pressure into account, you're going to find more places you can't take your mechs but can take tanks than the other way around.

In addition to the inherent mechanical inefficiency, you have a vastly worse target profile. Relative to mechs, tanks are essentially lying down, giving the smallest possible target profile for their size. This is also reflected in the higher ground pressure of something like a mech compared to a tank, which means they are more likely to get stuck in soft ground or be less capable in the kinds of rough terrain that they might theoretically be better at.

I haven't even gotten to the worst problem, that of the square cube law. Humanoid shapes are effective at the scale of humans. So this means powered armor is somewhat plausible but mechs aren't.

EDIT: This isn't directly relevant to my earlier points, but I just found this clip about AT-AT speed and thought it was an interesting issue. While the official speed of an AT-AT is fast enough to actually be effective as a combat vehicle, there is almost certainly no way this is actually physically possible given the model we see.

• I like your answer. But you seem to be focusing on humanoid shapes for walkers. If you consider four-, six- or eightfooted walkers, some of your points become less valid. – Burki Feb 11 '20 at 9:18
• "From a physical standpoint, it is inherently inefficient." It's a big engineering challenge, but not impossible. Energy is only wasted when turned to heat, else it's just converted: Lifting a leg creates potential energy, accelerating it creates kinetic energy, etc.. You could design a system with springs, masses and/or electromagnetic actuators that recover these energies. Hard, but not inherently impossible. – alain Feb 11 '20 at 11:53
• Your answer seems to be build on the premise that the "road" is flat and horizontal, maybe going downward, or only slightly upwards. Climbing (e.g.) a mountain while avoiding (or fighting) trees and rocks / boulders, makes the "bicycle" argument useless. The original question seems to be more about efficacy, rather than efficiency. – virolino Feb 11 '20 at 12:32
• "would you ever challenge a bicycle to a race on foot?" – YES – for example, a race up any hill which requires scrambling; a race with a significant sideways gradient; a race over boggy ground where wheels will easily sink; a race which includes hurdles; a race on paving slabs where you're not allowed to step on the cracks... legs are more effective on many annoying types of ground, but society has organised itself largely in places which have flat ground and roads because they make wheels more efficient, so we rarely have wars in those places. – Dan W Feb 11 '20 at 14:07
• > The amount of complex engineering that would have to go into this process is not worth the benefits. This seems to be strongly pegged to the limitations of current technology. For the purposes of world building, one could realistically assume that technology has advanced well enough to consider such things as "problem solved." – luis.espinal Feb 11 '20 at 21:12

## Yes!

This is a full functional John Deere walking tractor and tree harvester. On flat terrain (eg Hoth :)), wheels and treads may be more energy efficient, but in obstacle filled terrain, legged tanks may well be more useful. However, legs should be short and bent to keep the center of gravity low and the profile small, allowing the tank to not tip over and to take advantage of cover.

• While exceedingly awesome, the primary benefit of the Plustech Walking Tractor over tracks is minimal impact on the forest floor. This allows it to harvest deep in the forest without roads nor damage. Not a great concern in combat. – Schwern Feb 12 '20 at 2:05
• fantastic answer, thanks! – Fattie Feb 12 '20 at 13:14
• @Schwern Actually... You could make a case for legs allowing for "stealth tanks". Need to get forces in position w/out satellites/plans spotting you? Movement through thick forest on legs might be the answer. – aslum Feb 12 '20 at 17:30
• Not obvious from this picture is that this vehicle is incredibly slow. At the speed it moves, it's hard to argue that any kind of mech would have a real advantage over infantry. – Turksarama Feb 12 '20 at 22:36

# Surely...

everyone here has seen the utterly terrifying dog-like robots the US military is working on?

(Hilariously, they have more recently done stuff like paint them yellow and pink in a pathetic effort to tone down the "utterly terrifying" aspect.)

Are mechanical legged vehicles (or mechs) useful in combat in a near future scenario?

It's probably the single most intense desire of the US ground military at the moment.

You've completely hit the nail on the head.

If so where?

Every single region in which the US fights (ie, the overall planet other than I think Tasmania).

# 4 is easier than 2

4-legged death robots are incredibly easier to achieve than 2-legged ones.

Consider the case of Honda's totally amazing "Asimo" - while it's amazing (A) it can do nothing and (B) no matter how many decades they pour money in to it, it can do almost nothing else. Whereas, Boston Dynamics terror-dog makes huge advances every few months.

Four or six leg locomotion is incredibly easier than two (not easy, but easier). It's getting closer to being a solved problem.

• I can't say I agree with you about it being "utterly terrifying". BigDog isn't any more scary than their mule, cheetah, or wildcat robots, which weren't scary at all. They aren't cute and cuddly, but you wouldn't want a military robot to be that, would you? – computercarguy Feb 11 '20 at 18:31
• @user560822 put a gun on one of those and suddenly it is a terrifying war machine. They don't do that because when minimising collateral damage it is safer to have humans doing the shooting; but in a all out battle, a walking autonomous machine gun makes sense. – Davidmh Feb 12 '20 at 13:39
• but @Davidmh , people said exactly the same about the first cannons, drones, clubs, aircraft, etc. Note that you can take an elite commando out of action with small arms fire or a net. – Fattie Feb 12 '20 at 15:58
• Another key point is that these robots are not expensive to mass produce. It costs the US about 850,000-1,400,000 dollars to train, board, and pay an average soldier who is only really good for a few years of active service. In contrast, these 4 legged mechs typical cost less than a car. When you stop thinking of them as little tanks, and start thinking of them as expendable soldiers, their utility value goes through the roof. – Nosajimiki Feb 12 '20 at 20:34
• @user560822 they look less vulnerable than a squishy human. Small arms fire anywhere on the body will put a human out of action, this things can probably take a few bullets to the legs and body and still fire. – Davidmh Feb 13 '20 at 5:35

The top two answers are absolutely wrong!

This debate is always the same. And always you have some dense smartasses wobbling up with their "nu-uh wheels and tracks are always better!", which is about on the level of toddlers.

Yes, you will NEVER see Star-Wars style walkers, because they're impractical. and you will almost certainly not see any walker vs tank combat ever, because tanks win that easily.

Now, where WILL you see walkers? Simply any and everywhere where tanks are not viable.

• Urban combat (here small walkers, bigger than humans but still small enough for most urban openings and walkways. Aka you have a tank where you normally can't fit a tank.
• Mountainous terrain. Where tanks and generally any regular vehicles fail. Did you know that 21st century armies still have dedicated donkey handlers? And guidelines for horses? And what are horses but medieval walkers? So as soon as it becomes viable, they will be replaced (or at first supplemented) by walkers. However that as well will only happen within the narrow mission specifications they are used in today.
• +1 The example of horses and donkeys is a really good one. – Dohn Joe Feb 12 '20 at 12:46
• "here small walkers, bigger than humans but still small enough for most urban openings and walkways" AKA Power Armor. – nick012000 Feb 12 '20 at 13:06
• @nick012000 yeah, imho there isn't a clear distinction between a small walker and power armor (and you can argue that power armor itself is always just a minimum-sized walker) – Hobbamok Feb 12 '20 at 13:26
• @MorrisTheCat to your first objection, most images I have seen from urban warzones show collapsed buildings and rubble. I think a walker would have an advantage over wheeled vehicles. – craq Feb 13 '20 at 3:39
• Good answer, but you might add the jungle among the environments where legs are required. – FluidCode Feb 13 '20 at 23:55

Nailing my colours to the mast, I'd say that the answer is no; in the near future, there is no real use of a combat mech that can't be (better) met by a wheel or tracked vehicle. I was going to point to the fact that there ARE no mechs currently in development by defence industries, and that this is a sector which would be all over such a design if there was a benefit to it, but then I did a little extra research and found out that there is a Russian combat walker currently under development. I recommend you actually read the article as it covers a lot of the arguments for and against such a vehicle in a combat situation.

For my own part though, I've always been of a mind that walkers give you the same vulnerabilities that a human combatant has when it comes to legs; if you shoot at them and damage them, your mobility is screwed. Sure, they ride higher, giving the pilot better visibility, but that also comes at the cost of higher centre of gravity, meaning that balance is more of an issue for a vehicle you need to keep upright at all costs. The idea of walking over obstacles and the like that other vehicles can't get over does sound appealing, but then I came across another Russian invention...

Meet the Sherp, a Russian ATV that is designed to go almost anywhere. If you look them up you can see that these things can pretty much cover any terrain, still have reasonable visibility, and the low pressure tires make short work of pretty much anything that the walker above could get over, probably more. Sure, the wheels are a vulnerability just like legs, but you can still move one of these on 3 wheels (probably) and as the tyres are low pressure, they can take more damage before becoming inoperable.

Mount a 50 cal on the top of this thing and you have an interesting weapons platform that to my mind is cheaper to build, easier to develop as we've been using a lot of this tech for a century already, and (perhaps most importantly) is going to be a lot faster.

This is the final point I'd make in the argument between the two platforms - there is going to be a maximum speed possible for the walker due to the fact that you have joints moving backwards and forwards, rather than building up solid angular momentum in the same way a wheel can. This means that the legs actually have to surge energy in an alternating sequence instead of just letting an engine apply it constantly. This makes legs more complex, harder to maintain and puts a limit on how fast they can go by comparison to a wheel, which is also more efficient.

So, for mine, I don't think that combat walkers will be seen on battlefields in the near future, but if they are the only reason I can think of is to break the morale of the enemy and instill fear in them when they see a combat mech running at them, chain guns pointed their way.

• That thing will still flip on a deep slope or sufficiently difficult terrain. I kinda like EstimatorNoiseless's answer. – luis.espinal Feb 11 '20 at 21:13
• a Russian combat walker - I like the idea of Russian battle robots destroying the world, but that thing on the picture, is it even serious? – Headcrab Feb 12 '20 at 2:47
• "there is no real use of a combat mech that can't be (better) met by a wheel or tracked vehicle." yeah sure, because tracks work so well on mountain paths... – Hobbamok Feb 12 '20 at 10:19

Yes, but probably not the things you see in Star Wars. The most useful legged transport would be over very rough terrain where wheeled vehicles would not be feasible, and they would probably use 6 or 8 legs like a spider. This would likely be in hilly, heavily wooded, or mountainous terrain. They could be carrying supplies. They also could incorporate wheels if needed, when they reached roads or terrain that would accommodate wheels, and then switch back and forth like a Transformer. The AI that would be needed is probably available now, so by the time that these legged transports are needed that would not be a problem. Power supplies are currently a problem for this concept. They would probably need to incorporate their own power replenishing sources, such as rolling out solar arrays before their batteries are depleted.

Yes and no. Depends on your definition of combat really.

The Defense Advanced Research Projects Agency (DARPA) had the LS3 Legged Squad Support System project going for a while. The project is essentially the military version of Boston Dynamics' BigDog, and the goal was to serve as a mechanical pack mule for US infantry. The LS3 was used in combat exercises succesfully, and it served its purpose of resupplying units in placed where traditional vehicles had trouble reaching them. It performed admirably, was capable to follow soldiers through rough terrain, was capable to pick itself up when it fell. There was just one minor detail: it's LOUD. So loud in fact the USMC couldn't see themselves using it in battle.

In the end, the Achilles heel of BigDog was its gas-powered engine. Boston Dynamics also developed Spot, a lighter, silenter, electric-powered version of BigDog, but its carrying capacity wasn't sufficient. Fiction can help your case however. With the development of electric vehicles comes the development of battery technology, it's not hard to imagine that, in the near future, we'd have batteries good enough for a fully electric BigDog without sacrificing its capabilities.

Now here is where the yes and no applies.

As the name squad support system suggests, it is designed to carry equipment and supplies, like extra ammunition, heavy ordinance, medical supplies or whatever you deem useful to take in battle that you'd rather not carry yourself. While this can be used to support units in combat, this isn't quite a combat role.

Outside of that particular application, you could integrate support systems to it, like an antenna to serve as a walking comm link, or some kinda of radar or detection system to serve as a spotter for artillery or just to scout an area. Basically anything that you can bolt on a jeep, you could bolt on a BigDog and it would follow a squad just about anywhere.

You could even imagine a specialised version camouflaged as an actual dog and filled with spying equipment that you could send for reconnaissance as a less conspicuous alternative to a flying drone. And of course, in a pinch, you can always jury-rig it with explosives and kamikaze it against the nearest tank or entranched position.

Outside of combat it could be quite useful in disaster relief or rescue operations in mountains, forests, rubble, or basically anywhere where roads are impracticable (or anywhere without roads to begin with) as a much cheaper alternative to helicopters.

If you're looking for a more offensive role than that, don't. There isn't much of a use case for a mechanical mule with a canon. Anything it could do, a tank or UCAV could probably do better.

• I'd suggest that anything that is going to be shot at is probably better off with wheels or treads than legs, but if you need flexibility rather than durability or speed, a legged scout robot may well be more versatile than a wheeled one across different types of terrain. Wheels and treads tend to have difficulty with stairs for example. – Ruadhan Feb 11 '20 at 16:09

Nicholas Moran aka The Chieftain, former US Army tank commander and Wargaming America's historian, covered "Giant Stompy Things" (GST) vs tracked vehicles in his Q&A #7 at about the 10:47 mark.

His opinion was no, "mechs are stupidity inefficient designs" for reasons mostly covered in other answers but l'll reiterate. GSTs come out worse in height, weight, frontage, complexity, and ground pressure.

With any armored vehicle you want to get the most out of your weight. More bang per pound means more range, more speed, more maneuverability, less strain on the drivetrain, more capacity, and lower cost.

Modern armor are basically an armored box which makes them fairly compact. GSTs are not. All that extra surface area means more to armor which means more weight for the same volume or reducing the useful volume to compensate.

Modern armor have small frontal cross sections making them harder to hit. They concentrate their armor in the front; that's where they're likely to be shot at. GSTs have very large frontal cross sections making them easier to hit and requiring a greater percentage of its surface to be armored: more weight.

Ground pressure is a critical measure of the mobility of a vehicle, it's the vehicle's weight distributed over its contact with the ground. Riding a bicycle on sand illustrates this nicely. Tracks spread the weight of the vehicle over a large area consistently resulting in an 80 ton tank being able to go places much lighter wheeled vehicles cannot. GST must lift their feet off the ground increasing their ground pressure. This defeats the main touted benefit of a GST: mobility.

On a related note, a while back I calculated the weight of a Star Wars AT-SE walker using ground pressure. It came out heavy, but not unreasonable.

GST require multiple points of articulation per leg. Each must be powered. This is a maintenance nightmare, as well as a challenge to transmit power to all those joints. Those complex joints will all be exposed to incoming fire. All this means more weight. In contrast, modern tanks have an all-in-one engine/transmission power pack safely at the rear of the vehicle. Only the rear two wheels are powered. The tracks are relatively simple. Most maintenance can be done in the field.

The best way to protect a vehicle is for it not to be seen. Modern armor is low to the ground making it harder to spot. GSTs are tall making them easier to spot. Modern armor can hide behind terrain and take advantage of hull down positions to fight with only the turret exposed. GSTs can also use terrain, but being tall they will have less opportunities to do so.

What if you made a crawler, basically take a tank and replace the treads with stubby legs? This solves the surface area, frontal area, and height problems, but the ground pressure and complexity issues remain. It still has to lift its feet. Its legs still need extra motors and hydraulics adding weight and complexity.

Because these are essentially geometry issues, they remain no matter what sci-fi technologies you propose. Any advances in armor or power or propulsion or weaponry to make GSTs more feasible can be applied more efficiently to a tracked tank.

To sum up, for similar volume and protection, walkers are heavier, larger, taller, more complex, and less mobile than a modern tank.

Picture where an armed and armored horse would be useful in modern combat, and I think that will be 90% of your answer. Uneven yet solid terrain with good cover, likely for recon and small-scale engagements. Less useful than tracked vehicles where the ground is soft (mud, sand), probably get smoked in open terrain where wheels are faster than legs for the same mass (and missiles are faster yet).

Personally, I don't think we'll ever bother with legged vehicles or shooting platforms - there's a point past which legs don't scale very well (there's a reason elephants don't gallop). Instead, we'd build man-sized legged drones and 'bots for recon, support, and some close-in combat. Picture Boston Dynamics' Atlas or Spot, but armed.

Yes, in a combat sport.

Legged fighting machines are discussed so often because we find them so cool. So they will have a place in any scenario where coolness can take precedence over practicality. That won't be the battlefield, but it could certainly be a sports arena where we go to watch Pepsi Pummeler take on the Red Bull Brutalizer.

• Something like RCL? – Matthew Feb 14 '20 at 16:17

Two legs would be utterly pointless in ANY scenario I can imagine, inflexible, difficult to hide, single point of failure, ....

A spider tank could be fantastic. Redundancy, speed, power, all-terrain, etc.

A millipede-style tank might be even better:

I'm imagining Tiny, barely articulate legs with a single bend and very little internal control, maybe just a "Muscle" lever to operate the knee--all the hardware is safe inside the vehicle

• Under partial load you would only need a few legs and could be very quick and efficient.
• it could instantly scale to a full load by using more legs.
• tiny simple legs could be pulled into the body for safety
• New legs could be printed while moving to replace damaged ones.
• Would be amazingly sturdy and support any terrain
• Could probably walk upside-down on many surfaces due to the distributed footing and ability to apply opposing force.
• Very light "step", pressure per foot distributed over very large area.
• Very smooth travel.
• Could be adapted to nearly any vehicle. (You could probably put this kind of feet on a house and have something that could run down the freeway without disturbing the contents!)

Actually it would probably have a large load vs speed swing (Heavily loaded it would just be able to crawl) so maybe that house doesn't climb up on the freeway...but it would still look pretty cool..

• "Could probably walk upside-down on many surfaces" - what size are we talking about here? – Ruther Rendommeleigh Apr 1 at 15:14
• Size doesn’t matter too much, just weight per foot and the strength of the surface. If a surface can support a vehicle, then the vehicle could hang from the bottom assuming each foot can support its fraction of the weight with a little hook or angular pressure or suction... with a two legged robot each leg must support the entire vehicle in the area covered by one foot’s surface while it is stepping. I could see something the weight of a car walking right up and over a house without damaging it. – Bill K Apr 2 at 16:13

Ghost in the Shell may have the right idea... Tachikomas have four legs, but they also have wheels on the legs, which gives them some of the benefits of both. For level terrain, they roll like wheeled vehicles, bypassing the efficiency problems of walkers. For rough terrain, they can retract the wheels and gain the mobility advantage of walkers.

### It depends on the threat model

With drones and missiles becoming sufficiently smart and ubiquitous, heavy armor and a low profile will no longer be adequate protections for a manned ground vehicle. You'll need active defenses, at least some anti-air and electronic warfare capabilities and a versatile sensor and communications suite.

Even if you manage to make room for all of this (e.g. by replacing most of the crew with technology) in a traditional tank, positioning especially the defensive weapons in a way that doesn't restrict their arcs of effect or interfere with other systems is difficult, at least in my experience. This issue will be exacerbated if you need more than one offensive weapon for whatever reason.

For a vehicle favoring sensors, active defenses and utilizing a wide range of possible cover over heavy armor, the variable ground clearance and flexibility of legs (whether they're attached to feet or wheels) can be a major advantage.

Bonus points for a sectioned design that can take a penetrating hit to noncritical areas and remain somewhat operational.

(No, not humanoid. Funnily enough, aircraft-like designs - without the wings/rotors - seem to do well in simulations.)

### It depends on the tech level

Among the most often voiced concerns are complexity and energy efficiency, and they're 100% valid when talking about current day technology. However, complexity is relative, and what seems challenging today might be child's play in a couple of generations. And while e.g. the wheel is unlikely to significantly change in the future, mechanical legs and their control systems are potentially capable of feats that a car just can't replicate.

For example, imagine an AI that knew exactly where to step and how hard, could track any number of threats and optimize each movement to maximize cover against them, while moving across difficult terrain without slowing down. Put the same AI in a traditional vehicle and, well, 90% of its capabilites are wasted. But will we perfect a walker's balance, mobility and ease of repair before ground combat vehicles are made obsolete by some other technology? That's entirely up to the writer.

### It depends on the size

A roughly human-sized design is possible, although expensive, with current day technology and may be used to support infantry. More importantly though, if technology advances to the point where drones and anti-missile weapons are cheap and ubiquitous, you'll quickly find out that a lot of those weapons are also frighteningly effective against infantry, so it stands to reason that soldiers will mostly be replaced by (or at least operate exclusively in or near) highly mobile vehicles. It's possible, whether for ethical reasons or because your AI just isn't quite good enough, that those will still be operated by humans, and to keep that human in one piece, they can't be much smaller than a current day tank (though probably much lighter).

Scaling up from there heavily depends on what technologies emerge in what order, and has diminishing returns (so no "land battleships", ever), but realistically, the limit is determined as much by doctrine as by physics. For example, in one of my settings, making the walkers larger made them worse at combat, but able to carry bigger reactors, computers and sensors and thus more of a force multiplier.