Human sized war robots with legs, practical or not?

[Zor]

This is something i have been thinking about. We have long ago have came to the conclusion that Legged military vehicles (AKA Mecha) are just plain inpractical in real life as legs offer no advantage over treaded and wheeled vehicles. However i was thinking, what about robots on a smaller scale? Some of the problems of legs are considerably lessened on a smaller scale and on a smaller scale legs do have some rough terrain advantages. Even DARPA has shown some interest in legged robots, leading to the development and creation of BigDog. The question here is that could legged war robots be practical?

Zor

[Gunhead]

It depends. If you can produce warbots that are equal to or superior to humans as infantry at the same cost, it would be practical. The ways to go are really cheap easily replaced cannon fodder or extremely expensive but very effective.

The issues with any mechanical device are maintenance, how much and how often, initial costs, energy requirements and what is it capable of.

For example if you can produce, maintain and field an infantry bot that is roughly equal to a fit adult human in terms of mobility and weight, can perform complex tasks such as manning a checkpoint and it costs about ½ of what it would cost to train a human equivalent, it would be usefull.

Then there are several logistics issues to go with this that might make the infantry bot even more appealing such as standard parts, fuel etc.

-Gunhead

[Winston Blake]

Warbots bigger than humans are already in use (UCAVs) and development (UGCVs), so the robot part is fine. However, legs are probably always going to be for odd niche applications, like BigDog there. I can't actually see a reason why a BigDog couldn't have tracks, but if the tech gets worked out, I can imagine one of those popular little Packbots having four legs for scouting out particularly hard-to-roll-to places.

[Sikon]

For anyone has seen the replicator hordes of Stargate, imagine how the basic principle of a large number of small, cheap opponents could be dangerous if they had guns and grenades.

Small, cheap robots might be produced in large numbers, if the ratio of economic output to population rises in the future as it has over past decades and centuries, with much computer and robotics advancement.

In some cases, regular human soldiers may be a little like giant, expensive, poorly-armored mechs in comparison to them. Aside from training costs over the years that can amount to hundreds of thousands of dollars, it is worse to lose a human soldier than to lose a million dollars of equipment, at least if the nation is like the U.S. today.

As a random example, if one day mass-produced mini-robots cost the equivalent of $10,000 each and if $400 billion a year could be spent on them, that would be like a 1-billion-robot army being produced in 25 years.*

* Ongoing maintenance expense might be reduced if many are kept in storage until activation in event of approaching conflict.

If the small robots are around 10-kilograms each, each 100 tons of transport capacity amounts to 10,000 robots.

Here's one random example:
If a robot is 10 kg with an average density of 2 g/cm^3, one body shape would be its main body being 3 inches thick, 6 inches wide, and 15 inches long. In all-metric units, that's 8cm by 16cm by 40cm.

Such a small robot presents a small target profile to enemy soldiers, like the size of a person's hand rather than the size of a person. Like a rat hiding in grass, it isn't easily noticed from tens or hundreds of yards away.

Meanwhile, even ten kilograms of total weight is still enough for the robot to carry an accurate 1kg or 2kg computer-aimed gun lethal to regular human soldiers. It could also carry and toss explosives as large as regular grenades.

Since the robot's frontal profile is so small, like an elongated tank, optionally 1 or 2 kilograms of frontal armor could provide 10 or 20 mm of armor if made of steel. That becomes the equivalent of several tens of millimeters with more optimal materials for weight. At least today's rifle rounds penetrate much less.

The legs of insects allow them to go over common obstacles and variance in terrain that can be large in comparison to their own size. The same principle could be applicable with these robots. That's not just for macro-scale terrain like mountainsides and ravines but also for obstacles on their small size scale.

The mini-robots preferably have mobility against obstacles more like animals than like wheeled remote-controlled toy cars of today. The ideal is gripping feet, like insects and lizards that can climb walls.

The point here is not to suggest that hordes of small robots would replace other military forces. The principles of combined arms would still apply.

Among other limitations, if the robots are non-sapient, they can need sapient (e.g. human) commanders to deal with complicated situations, like if they ran into a scenario of needing to distinguish between combatants and civilians. But they could be an effective supplement to other military forces, if future technology allows them to have the appropriate performance without excessive capital cost.

[Lord of the Abyss]

If their AI is good enough and they can be made cheap enoguh I'd think them quite practical. And really, one follows from the other; a robot that can handle a war at human or near human levels should do even better in a nice, predictable factory; manufactured goods would be quite a bit cheaper I'd think.

One advantage a humanoid robot would have, is that we live in places and use equipment designed for humans. Buildings have human sized doors, human hand height doorknobs, stairs, and so on. Tanks are designed to be run by human shaped beings, guns meant to be held by human hands. All of that would make things less awkward for the robots, and simplify logistics.

[RedImperator]

I think if you're going to go with legs, something built more like a dog would be a better idea than a humanoid. For one thing, it's a lot harder to knock it off balance, and it can operate with one leg knocked out. For another, the target profile is a lot lower. If you want arms on it, you could just make them "centaurs" and add another set of limbs. There's no reason to stick with the tetrapod body plan for a combat robot.

[Master of Cards]

One advantage a humanoid robot would have, is that we live in places and use equipment designed for humans. Buildings have human sized doors, human hand height doorknobs, stairs, and so on. Tanks are designed to be run by human shaped beings, guns meant to be held by human hands. All of that would make things less awkward for the robots, and simplify logistics.

or you know, add a expanding arm and claw.

[Elaro]

What about stairs? Or other terrain optimized for humanoid use? Bipedal bots could find a use there. And It may not be so "niche" now that conflicts take on a more urban setting.

[Tasoth]

God, I can't remember the name of it, but they had a robot designed for used in scouting purposes for urban environments. I had a pair of tracks, a slew of various optical sensors on it and a mount to put a machine gun on that let it fire accurately for large bursts. Really, if robots are implemented in war, they're going to come out as mini tanks with either treads or wheels and probably some widget to let them take stairs, which many of the current ones can do. Making them smaller means they have a lower profile, making them harder to hit and easier to fit behind cover. Plus, if they're low, you can armor them like a turtle with a dome like armor top since most human targets would have to fire down on them.

[AidanMcfay]

Big Dog is fucking creepy. Oh and what is the noise going on while it walks outside? It sounds like a small toy plane engine. Dead give away if your trying to sneak into a building being guarded by Big Dogs.

[Setesh]

Big Dog is fucking creepy. Oh and what is the noise going on while it walks outside? It sounds like a small toy plane engine. Dead give away if your trying to sneak into a building being guarded by Big Dogs.

It was the test models engine, pretty much a weedwacker moter. (Hence why the indoor test with the power cable didn't make it.) Power is part of the problem with a legged robot, it takes more power than a tracked counterpart to run. Battery size vs reasonable weight limit is an ongoing battle. The test model's engine won't be used in a production model if it ever gets that far.

[Admiral Valdemar]

The energy density problem is solved with the radical new improvements in micro-turbines. You can run them off just about anything combustible and they are quiet, efficient and powerful, not to mention tiny. Those, combined with higher power batteries or fuel-cells, would make the power aspect nullified.

The bigger problem is intelligence, which could maybe be ironed out somewhat with pred-prey evolutionary programming.

Another advantage is that these bots would carry weapons no human could handle, so their custom firearms or grenades couldn't be used by the enemy against them.

To be honest, can you imagine how much better an army of thousands of these things in Iraq would be compared to the current few guys in an armoured car or APC constantly on the move to avoid fire?

The US is going in this direction, and Replicator style multi-purpose bots would be lethal against an army of humans.

[NecronLord]

Despite loving The Terminator, Necrons, Cybermen, and a dozen other humanoid robot groups as I do, you can do most fighting with cat or dog bots.

I can see some use in humanoid war-robots - Searching a captured desk's draws or typing instructions into computer would be hard for dogbots, and if you don't want humans in the combat zone at all, you'd probably need humanoid robots for that. In the same way, conversing with people is probably somewhat easier for a humanoid robot - especially if you can make it look human, a la Terminator - than it is for a dogbot.

But there's no pressing reason infantry robots need to be humanoid.

[dragon]

Hum that big dog was kind of interesting when it got kicked and it was able to right itself, but other than that legged vehicles are only good for certain roles such as climbing stairs and going into areas where wheel and tracks would have a hard time. I know most of the ground robots the military are working on are wheel and tracked based.

[Sarevok]

I dont think it would be very difficult to create an AI that can identify and attack human targets. The image recognition technology is already there. The big question is can AI developed in forseeable future can distinguish between civilians and soldiers. Without that capability military robots can only be used as berserking killer machines that eliminate everyone in a target zone. Not very useful in todays insurgency style warfares.

[Shroom Man 777]

Well, it could be simplified by programming the robot to purely fire at a person who is also discharging a firearm. If angry civilian mobs throw rocks at it and shit, it just fires off gas canisters. But if someone blasts away with an AK, then it returns fire. It'd have radars to detect incoming bullets and have split-second interception. Sniper fires a round? Triangulate, sight, shoot. Dead sniper. A bunch of guys spraying? Well, the robot fires a burst - one bullet for each target.

They'd walk slow and react slow to most thinggies, they'd be clunky and armored - but when bullets start flying, they can kill you in a microsecond.

They'd be autonomous, but can be remote controlled. For example, if Mohammad Derka Derka Al-Jihad causes shit and has to be killed, but he's unarmed, the robot obviously won't shoot him since the robot is programmed to fire only at targets who are also discharging weapons. So a controller can then remotely direct the robot and instruct him to kill Mohammad Derka, specifically just Derka, and no one else. You know, like a single-minded robot attack dog playing fetch the Muslim extremist with your titanium jaws.

Of course, they'd be useful only in combat-related duties. Mingling with the civilians ought to be a soldier's job, or else the civilians will start loathing the occupier and his robots. As for IED thinggies, we could have bomb-sniffing-dog-robot variants - remotely controlled, since I don't think robots would be smart enough to spot IEDs.

[Sarevok]

Well, it could be simplified by programming the robot to purely fire at a person who is also discharging a firearm. If angry civilian mobs throw rocks at it and shit, it just fires off gas canisters. But if someone blasts away with an AK, then it returns fire. It'd have radars to detect incoming bullets and have split-second interception. Sniper fires a round? Triangulate, sight, shoot. Dead sniper. A bunch of guys spraying? Well, the robot fires a burst - one bullet for each target.

There are two problems with this approach.

1 ) You are letting the enemy have the first shot. Once people figure out a robot's weak spots they are going to cripple it with a quick burst of small arms fire aimed at areas like the sensors, never mind being hit by heavier weapons like grenades or rockets which will destroy a robot regardless of where it hits. The robot will have no chance to defend against such an attack since by the time it register hits and decides to engage the enemy it is dead or crippled.

2 ) How do can a robot reliably detect weapon fire and track it's source ? Attributing a loud boom to a specific gun carrying invidiual and tracking him is not as easy as it appears.

[Shroom Man 777]

There are two problems with this approach.

1 ) You are letting the enemy have the first shot. Once people figure out a robot's weak spots they are going to cripple it with a quick burst of small arms fire aimed at areas like the sensors, never mind being hit by heavier weapons like grenades or rockets which will destroy a robot regardless of where it hits. The robot will have no chance to defend against such an attack since by the time it register hits and decides to engage the enemy it is dead or crippled.

Simple. Stick the cameras under bullet proof glass. It'd take a marksman to hit those sensors anyway, and most insurgents are decisively not marksmen. And even if there was a sniper, he'd have to decommission all the sensors at once. The robot will have multiple sensors. And before the bullet even touches it, it's radars will track the trajectory and the robot will be well on its way pointing its gun at the exact location of the sniper/insurgent.

As for RPGs, the robot will not be as big as a car, so an RPG hit will be very much unlikely or at least difficult. And, certain interception systems could be used. Track rocket in radar, point gun, shoot down rocket. Or we could have that ARENA system the Russians have - detonate the RPG by launching a mini-missile at it. Sure, it might cause a lot of splash damage, which is bad, but if the robot gets hit, there will still be an explosion that will hurt people. *shrugs*

2 ) How do can a robot reliably detect weapon fire and track it's source ? Attributing a loud boom to a specific gun carrying invidiual and tracking him is not as easy as it appears.

Radar? I think it's feasible. Of course, I could be wrong. Use the radar to track the bullet trajectory, as well as auditory sensors hearing the weapon discharge, opticals seeing the muzzle flash.

[Larz]

I can see some use in humanoid war-robots - Searching a captured desk's draws or typing instructions into computer would be hard for dogbots, and if you don't want humans in the combat zone at all, you'd probably need humanoid robots for that. In the same way, conversing with people is probably somewhat easier for a humanoid robot - especially if you can make it look human, a la Terminator - than it is for a dogbot.

Or you could just put robotic hands with a visual sensor on an extending stalk/arm.

As for the conversing aspect, why not just send a human? To me it would seem like a waste of space to give it the AI needed to carry on a conversation with a human. Really, if the situation where so dangerous that sending a real human would be too much of risk, I think that having to talk to an armored turtle would be the least of a persons stressors.

With the right tools I could see the use of the turtle/spider design of a military grade robot, but that bipedel design just seems far too silly. It's easy to knock over; has all the platform limits that a human has (it would need to brace itself to use high recoils weapons, it wouldn't have the same sheer carrying ability as that would mean more stress put upon just the two legs versus tracks or many legs, etc.); would more than likely be far harder to repair/maintain its two complex legs as well as they would require more maintenence as they would undoubtably have to withstand more stress per leg than a track/multi-legged system; and lastly as it has been pointed out bipedal designs are tall, easier to hit targets.

I don't really think there are any real good points for using a biped design that a tracked/multi-legged/mixed design couldn't accomplish far better, easier, and more cost effective other than for the whole "ZOMG, Terminators r s0 k3wl !!!!11!!shift+1!" reason.

[NecronLord]

As for the conversing aspect, why not just send a human? To me it would seem like a waste of space to give it the AI needed to carry on a conversation with a human. Really, if the situation where so dangerous that sending a real human would be too much of risk, I think that having to talk to an armored turtle would be the least of a persons stressors.

If you're going for an all-robot army, then you're likely too paranoid to risk humans without exceptional cause anyway.

[Larz]

If you're going for an all-robot army, then you're likely too paranoid to risk humans without exceptional cause anyway.

Still doesn't provide any worthwhile reason to make a bipedal human like robot that is at the same time capable of being used as a military unit. If your that paranoid and really that concerned about having a person talking to a person robot than just make a special relations andriod but doen't even bother trying to outfit it to double as a military driod for all the reasons listed before as well as one last one I thought of: if one leg on a biped goes out (damaged, stops working, etc) then all you have is an expensive, armed warbot thats as mobile as a stone.

[Jadeite]

2 ) How do can a robot reliably detect weapon fire and track it's source ? Attributing a loud boom to a specific gun carrying invidiual and tracking him is not as easy as it appears.

Radar? I think it's feasible. Of course, I could be wrong. Use the radar to track the bullet trajectory, as well as auditory sensors hearing the weapon discharge, opticals seeing the muzzle flash.

There's a concept called Boomerang that uses an array of microphones to triangulate the sounds of a gunshot toward the vehicle. It has a computer to filter out other noises, and a display that tells the crew the range, bearing, and elevation of where the shot came from.