Boston Robots

Yeah, but when you use the horse as stress relief you all have to share. With the robot you can either issue each soldier their own hole, or at least remove it for thorough cleaning if you can only issue one per robot.

Eventually they will do the majority of real fighting in wars. Robotics is going to change a lot of things in the next 40 years

Mule:

...

Not just
but

What is it with shooting horses that break their legs anyway? Is it just a Western film meme? People break their legs all the time and after a month or two being in a cast walk on them just fine. Is it something about horses anatomy that makes it impossible?
Sage because offtopic.

Up until extremely recent advances in material sciences, there wasn't a cast material strong enough to be used for a horse's leg, since you're talking huge differences in bodyweight. That meant either burning truckloads of cash by caring for it for that month or two with a suspension harness and other shit, or cutting your losses and shooting it in the head.

Horse leg bones more often than not shatter and can't go back together.
Also the muscles are in the upper leg so there is barely anything to support even simple fractures below the knee, can lead to the fracture getting far worse and more serious injuries in the future, like a weak leg rebreaking and literally get ripped apart when that happens.

The lower part of horse's leg basically corresponds to what would be fingers/toes on a human or other mammal. What you might look at on a horse and think is its knee is actually its wrist anatomically speaking.

This area is mostly comprised of tendons and bone and can be very fragile and when broken it takes a long time to heal.

Attached: serveimage.jpg (600x756, 118.5K)

MOUT assault. This type of combat is personal intensive and dangerous even against low tier opposition. Walking into room where can be full auto AK welding terrorist is never a safe enterprise. With robots you risk just robots so no real risk. Ideal gear to augment SWAT squads when bump fire stocks confiscation time will come.
hooktube.com/watch?v=Ul04Is5K3nE

Attached: tyuty.jpg (970x545, 202K)

When a horse can cope with the recoil, noise, heat, spent brass etc from firing two rotary barrel miniguns whilst also ignoring small arms fire hitting it then it has the same potential. Imagine pic related if it was built by DARPA rather than some special effects guys.

Attached: T-1 Battlefield Robot Hunter-Killer.jpg (575x377, 115.24K)

They dont have appropriate power source and will not have for long time. It's DARPA shekel grabbing.

>tanks are impossible there is no appropriate power source and will not be for a long time. Tanks are (((armored corps))) shekel grabbing.

Except motors were constantly improving at a fast rate while batteries still have shit energy density (not even a proper truck is viable now and cars have somewhat limited range, imagine a walking shit with way less efficient movement mechanic than a spinning wheel).

Considering the lowest cost estimate for a battlefield version is 100,000 while a wheeled robot of the same size costs 3,000 the answer is never ever.

Boston is doing good research but it won't be viable until humanity stops feeding the sub races.

Totally unbiased analysis from you man, how bout.

Mule:

Attached: cdc.jpg (546x366, 53.05K)

Trust canuck to make abhorrent post.

Attached: 14725029756030.jpg (226x240, 17.68K)

It runs on an internal combustion engine, what are you talking about m8.

You know you can put ICE on the robot?
Well even worse for you, mister terrorist, not only robot will rip your anus apart, you would be also suffocating from the engine exhaust when it enters your hideout, ha ha ha :^)

BTW batteries or even better, methanol powered fuel cells have enough energy to easily provide robot with 1-2 hours of walking. So it is already enough for the mission:
1. Roll up to terrorist house in the APCs
2. Deploy robots
3. Assault house
4. Collect profits
5. Recall robots to the APCs, swap fuel cell cartridges, roll to another terrorist house.

Attached: 123123.jpg (800x531, 382.9K)

I just want riot bolis riding rhino size armored bots

Most of those are fixed with proper AI.

Niggers are cheaper than motorized farming equipment, but still everyone uses the second because it's more effective for the specific task.

It ran on a petrol engine right up until the point they realized it was noisy and can't follow indoors, every subsequent version of Boston Dynamics' robots have been electric.

Which would cost trillions to develop. And on each bot you'd have to put a fucktonne of sensors, raising per unit price into the millions.

Fucking wrong. People use machines because niggers are lazy and subhuman.

The bots already have sensors though, the spinning things on the top/front.
The AI is already developed, that's why it can keep balance when pushed and redefine a path.
Do you really believe this?

Do you realize what a general AI even approaching a mules intelligence would cost?

The thing could predict markets and let any company that owns it basically rule the world.

Oh and Boston Dynamics quadrupeds run on preprogrammed paths. It can react to anything that distubs it from this path in extremely controlled circumstances, but it can't decide on a new path. For example the door opening bot can only open that one door.

What this leaf said. there's a huge difference between developing an AI that can take steps forward through an environment without falling over, and an AI that can make intelligent decisions about moving through said environment like a real animal. The closest you could get is a legit animal brain jacked into the automaton (which has been done on small scale)

What he said ^

In terms of AI we're basically struggling with cockroach-level.

So … RL Servitors? Would it not be easier to have each one be remote controlled by a meat operator? Or if we're sticking with putting an organic brain into the machine then why not focus on prosthetic enhancements for the existing soldiers.

Cockroaches can reroute paths.
AI for it is not hard and already exists. I think that you fell for the meme that AI is some kind of sci-fi unachievable tech. It doesn't need to simulate a mule, it just needs to be able to follow someone or walk in a direction.

Hopefully waifu bots.

Attached: 0193d9bb4f62160e2953c8e0aaee4e3d69bc97f76d979abef3ba2566048ded6a.png (1123x1391, 369.78K)

So can humans, but that doesn't mean that an AI that can mimic 1 thing a human can do is as intelligent as a human.

Cockroaches can do a fucktonne more than just reroute paths. As an example they can also learn, which even the best computer program can not do. In fact despite not being able to hold as much information in their wee brains, they can learn as fast as a dog. Their brains operate in a range of 1-100 petaflop, estimated, whereas the fastest supercomputer ever devised has 93 petaflops. Safe to say putting a server farm in a bigdog, and somehow make it as CHEAP as a cockroach, is not going to be possible for the forseeable future.

The closest we have come is a program starting out with parts that let it do a LOT of things, then self-deleting parts which it doesn't need. In terms of self replication, software (quines) are at the level of pre-life abiotic soup, which included complex amino acid and RNA like proteins. It's not even proper bacterial cloning yet. They can't do horizontal gene transfers, such as swapping and integrating data between different programs.

By the way a Big Dog might be able to walk in a straight line or follow someone across a parking lot, but… just walking through a forest a Big Dog will either have to have a library of densities and tensile strengths for every object in that forest, as well as the ground itself to avoid falling…. or it will have to develop the ability to ESTIMATE these things from incomplete datasets. It would have to either have a map of said forest to plot paths around trees, copses, bushes… or it would have to ESTIMATE when the tree, bush, whatever obstacle will stop and it can rejoin whatever its following. Do you realize what that would take?

There are plenty of algorithms that can simulate learning.
Oh wait, you're going to define "learn" in as "is a general intelligence". Well you've got me there, we don't have a single computer AI that can do everything. Luckily Big Dog doesn't need to be able to do everything.

Comparing meat brains to silicon ones is a false equivalence. A cockroach can't learn chess in 4 hours and then go on to defeat grand masters. It can't generate the parse tree of an english sentence. It can't find pi to a trillion digits. Meat brains are general and made for survival, computer algorithms are idiot savants. I'm talking to too. Unless the goal of the algorithm/computer is to directly emulate or simulate a meat brain, rather than simulate/emulate a task preformed by a meat brain, comparing the two just makes you look ignorant about both.

Give me an example. None of what you're talking about resembles anything modern outside of video game AI.

But they can, and the tech isn't even state of the art at this point.

Has anyone talked about the possible logistics rather than combat use at first? is quite right. Has anyone seen that vi of a BD robot doing backflips?
Won't do much if they get jammed like Iran did to the US drone lol

Is there a single program that can write, on its own and in response to a problem facing it, a functional program it has never been in contact with before? Is there even a single program that can scan another program, then copypaste the parts of the other program that help it to perform a given task?

Simulating intelligence in controlled environments doesn't translate into real world capability. The best military AI we have are on UCAVs, because air is a relatively obstacle-less medium, and even those AIs make retarded mistakes all the time (re: iran).

Are you talking about the IBM computer that "learned" chess? Because it didn't learn chess, it knew all of the rules of chess already, it was programmed with them. As for learning to play, the programmer gave it five million opening positions. This is exactly the kind of coaxing that every modern "AI" needs, none of them are actually intelligent.

Talks about their benefits, if they existed, which they fucking don't. Jesus Christ.

No you moron, path prediction, based on understanding the human its following and where they're likely to go if it ever loses sight of them.

Where's the pussy AI, DARPA

Hey, I was right, so I'll just reiterate what I already said. What you are asking for here is a general artificial intelligence, which doesn't exist.

No, Deep Blue isn't anywhere close to modern AI. I was talking about AlphaZero, which was given the rules of chess and mastered it by playing games against itself. It did the same with Go and Shogi.
I know Deep Blue didn't learn chess, but I don't see how being given the rules is an indictment of any Chess AI. Are you saying that in order for an AI to be considered "intelligent" it needs to be able to extract the rules of a system by demonstration or by being told them in plain English rather than by a programmer?
If you want AI that can understand human speech you can dive into the field of Computational Linguistics, in particular Natural Language Processing.
If you want AI that can learn to preform actions by observing there's Baxter made by the company "Rethink Robotics" which is a complete package, robot and AI.
If you want AI that can work out the rules of a system by observing and interaction, you'll want to look into "Cooperative Inverse Reinforcement Learning".
However I don't think that any of the above is necessary for Big Dog to be useful in the field. I was just bringing up AlphaZero to show that comparing an organic brain to a mechanical one in the way you did isn't a useful comparison. AlphaZero is pretty good at board games, but outside of that it's retarded.

Here's an excerpt from page three of the paper

Sorry, I misunderstood.
Predicting where humans will go isn't unheard of. Here's an excerpt from Stanford AI research paper
with an average accuracy of 93% supposing 3 km^2 of uncertainty.
That paper is "Learning to predict human behaviour in crowded scenes" by Alexandre Alahi, Vignesh Ramanathan, Kratarth Goel, Alexandre Robicquet, Amir Abbas Sadeghian, Li Fei-Fei, Silvio Savarese.
That section is just a summary of the paper the [1] is referencing, which is "Limits of Predictability in Human Mobility" by Chaoming Song, Zehui Qu, Nicholas Blumm, Albert-László Barabási.

OK.

I'm sorry but that's not intelligent, it's just a brute force attempt to solve chess. It's like saying a password cracker "learned" a password by simply trying different variations until it got one right.

How about you look into it some more.
jkoba.net/jkoba-wiki/index.php?The Microbial Genetic Algorithm
It's not doing it on its own, it's literally programmed to do so.

Very nice. This paper is about following a regular person for awhile, learning their patterns, and then using those to predict their position. How will you apply this to the battlefield? How will you observe the soldier? Who will do the number crunching? How do you eliminate the 7% where it guesses totally wrong, and narrow the three square kilometers to three square meters when it guesses "right"?

CLANG

You don't need the thing to be able to predict paths, just being able to walk in the direction it's commanded to walk in to be able to be used as a vehicle, the AI is only needed to be able to not fall down, which it is already capable of doing. As for unmanned, just lead it with a rope, or attach some kind of emitter to the soldier that is supposed to be following and make it walk in it's direction while teaching the soldier that it has to keep a reasonable pace to keep the robot from losing him.

I originally proposed it as a replacement for cavalry, and the bot has all the capabilities to do that.

Except for the fact that they are inferior to small offroad vehicles in every single way. There's a healthy debate about whether they're good enough to replace mules with good points on both sides, a horse is another ballgame entirely. the thing with big dog is that it's allowed to fail, so long as the failure doesn't end the mission its on. If a horse stumbled to the extent and frequency that big dog does it would simply be unsafe to ride. and on a mechanical side horse legs are very efficient, far more so than big dogs. Boston Dynamics would need a totally new design to even have chances at making a mechano-cavalry, and a higher budget too. I thought you people liked horses though?

I never redefined artificial intelligence, it's just become clear we were both using different definitions from the beginning. You're definition of artificial intelligence is General Artificial Intelligence. If it isn't general, then to you it isn't intelligent.
The definition used in AI research and the one I am using and have been using is that AI perceives it's environment and preforms actions to maximize it's chance of achieving a goal.

The learning method may not be evidence of critical thinking but the end result is intelligent and effective. "Intelligent" in this case meaning it perceives it's environment and does a good job at preforming actions to maximize it's chance of achieving it's goal of winning the game.

How else are AI developers supposed to develop AI?

That would be up to the people developing the AI if they choose to use this research.
I won't. I'll be sitting here arguing about AI on Zig Forums. If you mean "you" as in something or someone else then be more precise.
Some general ways of getting a computer to perceive a soldier would be cameras, laser scanners, drones, short distance radio transmitters, and GPS trackers. Many of these could be done while training the AI so that the ones that work better in the field could be used in place of the ones that don't.
A different computer ahead of time. The robot will then use the trained model. Have you ever written software that uses AI?
It doesn't have to be perfect, just as good as or better than the alternative.
It wouldn't be making predictions as far into the future as what you is seen in the papers, which will narrow the distance. I'm guessing worst case is it looses sight of someone for a minute or two, and the goal is it gets somewhere where it is close enough to reestablish contact. It shouldn't be following too far behind anyway.

I was using the two papers mainly to establish that predicting human movements is possible, though I will agree that viability in this area is still in question. But as points out there are other methods. Another is a hybrid between camera tracking and a transmitter. Once it looses sight it pings a device on the soldier, and that device pings back the relative location. The robot can then navigate from point A to point B while it searches for the soldier and avoids obstacles. This information is encrypted of course.
I don't anticipate the training required to get a soldier used to dealing with a robot will be much different than the training required to get it used to dealing with a horse or mule used in the same situation, that is, carrying gear. I'm not of the opinion that it would be viable for cavalry on the modern battlefield.

Though I can't believe nobody has mentioned the true champion of this debate, the WWII era Jeep.

I stand corrected.

What about vehicles with legs that have wheels at the ends? That is certainly superior to wheels fixed on an axis, though this is not what the thread is about so sage

Oh, there are batteries with good energy densites that don't rely on a limited resource (Aluminium air batteries) downside is they're not rechargeable and need to be recycled.

Honestly, if you're going for legged vehicles you may as well just do an exosuit. A rubber tracked vehicle with adjustable suspension would achieve everything you need from legged wheels and more. im sorry, america isn't called brittania and knightmare frames don't real yet

Are exosuits plausible in the near future?

Exosuits are entirely possible now, they just aren't very good on power. Here's a discontinued american project by raytheon for you to look at. The americans canceled the project due to it being both slow, and having low deployable range. They're looking into a lighter stamina improving frame rather than a full suit. I still think the full suit would be interesting for AT teams and Machinegunners.

sage because now im extra offtopic.

Sick of arguing this dumb intelligence point you are so obviously wrong on, regardless on how intelligent your robot is the fact remains it's less intelligent than an insect, and you won't be putting a 10km^2 supercomputer in it.
Call me when the bigdog costs $100-1000, the factory for it is just two other donkeys, it can self-repair, it can feed and lubricate itself from the environment, and it can operate for more than 20 miles before having to RTB.
Until then there's not much to discuss.

The only real independent exosuit is the HAL5, DARPA copied the leg parts for some ruck carrier. Regardless, it's plausible in the now-to-2050 range for sure.

Tractors don't mug and rape your daughter.

You cannot flay a truck into working

niggers cant get as much done as fast as a tractor can, and i doubt youd be saving all that much after housing and feeding the niggers

Niggers are more expensive than farm equipment. Capital costs are greater because you need shit ton of niggers to do the job one tractor does. Human food is more expensive than motor fuel counting by the amount of work done (even if you go by the cheapest bottom of the barrel food). Not using slave labor as a substitute for machine labor is simply cheaper. Same reason why motor carriages had replaced horses everywhere: business only cares about value, and motor vehicles just have better value.

To paraphrase the last part. If it was cheaper to ship 100 000 ton freight by horses than by locomotive, then everyone would use horses for shipping.

I know you as a person would care if your internet purchase came sooner. But businesses tend not to care if their million tons of iron ore shipment is going to take a few months. They just want it shipped cheaply. Same idea with using beasts of burden for labor.

I can guarantee you that every production vehicle they make will either have conventional power plant or a fuel cell. Batteries just ain't gonna cut it in the field.

It could predict markets to the same extent a stock trader could. Probably much less. Shit like that is extremely chaotic and turbulent. It is somewhat deterministic, but you will not be able to leverage it to make an accurate prediction. Exact state of the system determines its exact state in the future, but approximate state of the system determines nothing.

I agree, but that's why the US army doesn't want it. no matter which power system they choose it sucks at something too much to see use. the only thing i'd be curious about is hydrogen fuel cells, but then im sure even that has issues.

Would you protect her from being bullied by Boston Dynamics' employees, Strelok?

I.e. no better than chance?

...

And this is how I know you're just blowing steam about things you know nothing about. You CANNOT brute-force chess, the decision trees balloon into the hundreds of millions after less than a dozen moves.

That's still trivial for a computer.

Why not just make the Half-Life 2 Hunter?

Attached: ClipboardImage.png (484x641, 164.61K)

It cannot store every board state. It is much more efficient to train it in another way. There are some good videos on YouTube explaining how it works.

SOON

Attached: 1444054995724.gif (364x243, 1008.14K)

It doesn't store all board states, just the challenges and responses that lead to victory. And a brute attack is exactly what the chess program did, it brute forced 2.5 billion moves before it discovered enough tactics to put it ahead.
The original claim is that it LEARNED chess. As in someone without even knowing the game existed, the program taught itself how the game works.
Instead the reality is the program was given the rules on how the game works, and then playing against itself millions of times managed to derive enough winning strategies to beat a previous computer (that was nerfed).

I like how the EMPs were the only weapon against the machines, but somehow nuking them had no effect.

I don't see anything unreasonable here

Way to waste a get on such a pathetic low-tier post.

Quads are not GETS

Yes they are. Even 12345 is a get. You're confusing dubs with gets.

Attached: th.jpg (474x472, 28.09K)

No fag. At least five repeating numbers. Anything lower is not notable. Quads are a one in a thousand, that means one every two or three full threads.

Three threads here are about 100 posts lol. Go back to kinder Zig Forums or /b/ or whatever.

...

>>>/4chan/

I'm not even going to tell you to lurk, just go back. Pic related is you.

Attached: PHOTO_20160730_205348.jpg (2560x1440, 1.15M)

You monster.

They won't. Google/Alphabet divested from the company because DARPA funding dried up. DARPA funding dried up because, although Boston Dynamics largely achieved their goal of solving the issue of locomotion, nobody solved the issues of power generation and energy storage.

Internal combustion engines aren't efficient enough to overcome the issue of fuel weight/cargo weight ratio (the amount of fuel needed for operational ranges nearly matches or even exceeds the robot's carrying capacity) and battery technology has been stagnang in terms of energy/weight for decades withoht any signs of a breakthrough on the horizon. Robots like Big Dog and concepts like power armor will never become a reality unless fantasy tech such as backpack sized fusion reactors somehow become viable by the laws of physics.

The most interesting thing to think about in relation to the shortcomings of these robots is just how energy efficient humans and animals are when it comes to locomotion. 1 gallon of gasoline contains roughly 32,000 kcal. This robot can barely travel a mile with that gallon of fuel while a pack mule can march 30+ miles on just a few thousand kcal of hay or other forage per day.

No way my dudes. What we need to be investing in is war tortoises.

Attached: A._gigantea_Aldabra_Giant_Tortoise.jpg (3264x2448, 6.03M)

I want to fuck that robot.

Soon.

Attached: robot_sex.jpg (580x385 920.79 KB, 40.62K)

So what you're saying is that instead of robotics we should be investing in genetically engineered/purpose built animals to do the same jobs and revive zoological warfare? Because I could get behind that, if nothing else the production costs are going to be a hell of a lot lower - if you make sure to feed them and don't mix any Panda into the mix they'll produce new units all by themselves every year.

Realdoll et al would do well to follow this advice as well.

Attached: Gengineered.jpg (1920x989, 264.57K)

Before you get too excited, this is just an art project. It still looks cool.

Sure, so long as she learns to make some pancakes, once in a while.

These gizmos are budget sinks and nothing more.

Their entire reason of existence is that the techno-fetishist US brass is unwilling to lighten infantry gear yet willing to spend billions on exoskeleton and robomule research because the troopers' knees get busted after six months just from carrying all that trash.
Which is an excuse to procure even more trash, the lobbyists need the money.

You forget you can also eat a Mule in an emergency.

PRECISELY

Or provide a mix of the two, which leverages best aspects of technology (control) with best aspects of biology (energy management).

Basically a cloned mule with a brain implant telling it where to go. We already cloned a sheep, cloning a mule should be easier. And we already have brain implants that can force even humans where to go.

Why restrict the new zoological components to carrying freight? It can't be too hard to breed a few new species of predator to unleash on enemy armies. Start with an unimproved Grizzly Bear, engineer its offspring for larger muscle mass & efficiency, more efficient metabolism, and work out a control system that's designed to be as simple as possible (it doesn't need to do much more than give direction for movement and have a 'docile/KILL!' switch). You can airdrop a few hundred to a few thousand of them into the enemies rear areas and have them head for production/logistical targets before you hit the psycho switch. You've then got at least a third of a tonne of very angry, very well designed killing machine tearing through a poorly defended target. It's probably going to take a lot of (panicked) intermediate calibre rifle fire to bring one down as well.

Attached: dismar5.jpg (1920x1080, 85.7K)

What would happen if you jacked a GPS system into a bear's brain? I think the best solution would be to wire some kind of a sensor to the bear that forces it to wander towards the target, but at a reasonable pace (so that it won't kill itself due to exhaustion), and also it shouldn't interfere with its natural pathfinding capabilities (lest it will run off of a cliff). Sounds like a difficult task to accomplish with a dumb animal.

Fair enough, just drop a few thousand bears into the terrain surrounding the target, give them a shot of bear PCP, and let nature take its course.