DARPA's headless Cheetah robot can run faster than you

Developed by Boston Dynamics (previously), Cheetah can run 18 MPH, needs neither to see nor scent you to know where you are hiding (because you're screaming), but is fortunately not aware that it is on a treadmill.


  1. Ahhhhhh, the first thing I think of is that nightmare killer hound robot from Fahrenheit 451.

  2. I’m glad you posted this. 

    I’ve always wanted to know the dynamics of testing something’s speed utilizing a treadmill. 

    Yes, the belt is moving at 18 MPH, the thing just has to lift it’s legs at a certain speed, is it comparative? 

    There is also no forward momentum, no push. I’m not an engineer, but I have run on a treadmill and I have run on  hard surfaces. I was told by a trainer that running on ground is like having ropes tied around your ankles from in front and pulling on them. Not so on a treadmill. 

    True, that argument may not make sense, but switch to rock climbing. What if the wall was moving down, and all you had to do was move your hands and feet to match the speed of the surface, you would never have to literally pull yourself up. It’s the same on a treadmill, you don’t have to pull yourself forward. Therefor, a lot less energy. 

    I mention this for one reason, I believe this is how some cars and their top speeds and MPG’s are tested, on a dynamo or something like that, there’s no forward momentum. 

    Basically, I don’t believe this kind of measurement.

    1. you fail to take into account the backwards momentum imparted by the belt to the robot/car.  If you stand on a treadmill and simply hop straight up into the air, when you come down onto the moving belt, your feet will be thrown out from under you.  To avoid this, you must jump ‘forward’, to counter that force than thereby land at a position on the belt AHEAD of your current center of gravity, so as you hit the belt, your feet will be brought back under you.

      In short, this sort of test works just fine.  There may be fine grained differences between staying stationary on a moving surface and moving forward on a stationary surface, but for purposes of judging the speed of linear motion, it’s perfectly valid.

      1. I halfway agree with you, it’s more about finding a new balance, a treadmill forces you to lean forward, to find a new center of balance.  

        For myself, the other proof is that different muscles hurt. My calves hurt good after a great workout on a treadmill, my quads hurt good after running on a flat road. That has to mean something?

        1.  A lot of it is to do with the lack of air resistance on a treadmill. That and the road is *hard* :)

        2. It isn’t about finding a new center of balance at all.  It’s about countering the momentum imparted to your body.  There is some level of angular momentum imparted because your center of gravity is well above the point of application, but by and large it is linear momentum, and the better your balance the less angular and more linear that momentum becomes.

      2. [replied to wrong person. Damn it, and now I lost the entire post, because Disqus’s stupid box arbitrarily loses focus, and so I failed to copy my text….]

    2. I think that the amount of friction involved — that thing must weigh a lot — would mean it must be moving close to the treadmill’s speed. Unlike a human on a treadmill, it’s not changing how it runs to stay in place with the minimum of effort. It’s just stabbing away at the ground the way it normally would.

      But your point’s a good one!

      Also, the plane takes off.

      1. Earth rotates at just over 1000 miles per hour. 
        According to your theory just hopping up in one place would be one of the fastest ways to travel. 
        If you want to go west. 
        (being facetious) 

        No really, the ground moving beneath you is identical to you moving over ground. 
        If it suddenly stopped running, the automated monstrosity would hit the conveyor belt and flip. 
        The difference really is just the lack of wind. 
        A plane takes off from airflow,  not ground speed.  Those just typically pretty much coincide. 


    3. No. Einstein would say that if you’re in an open-bottomed box with the ground moving beneath you, you would have absolutely no way of knowing whether the box is stationary on top of a treadmill, or the box is moving on stationary ground. It’s exactly the same from the point of view of physics.

      Air resistance is a valid point, but this is equally fixable. A fan in front of the treadmill, programmed to push the air at the same speed as the treadmill, will solve this.

      If you’re running with your eyes closed and the wind in your face, you would have absolutely no way of knowing whether you’re running forward on flat ground or whether you’re on a treadmill with a fan in your face. It’s exactly the same.

      1. Would he say that before or after he started screaming about being trapped in a goddamn box?

        I shall call this Einstein’s Uncertainty Principle.

      2. It may be the same _while_ you are running, but in getting there, there is a lot of difference. 

        On a treadmill: ground suddenly starts moving and you have to start walking or fall over (or keep your balance by straining and falling off the edge of the tredmill)
        On the ground: you start moving yourself and have to invest the energy to accelerate your body mass. That energy is not needed on the tredmill.

        So the point is: is the cheetah strong enough to accelerate his body up to up to 18 MPH? I think he probably is, because a machine can take as long as it needs to gain enough speed.

    4. There’s probably a small degree of difference, but I’m pretty sure that treadmill and wind-tunnel tests are engineered to get as close as possible to what is needed to measure the properties that are being measured.
      Thinking of your rock-climbing example, I’m picturing you on the low end of the deck of the Titanic as it up-ends and starts sliding in to the water. All you have to do is maintain the same postion in relative height by “climbing” the sinking object to stay above the water. You think that would take a *lot* less energy? I’m thinking you wouldn’t notice any advantage at all compared to actually climbing a vertical object.

    5. I wrote another long reply, but Disqus ate it. Short version: Running on a treadmill is exactly like running on a moving train. (A very smooth-moving train with no bumps or anything… Actually, that’s like a big cruise ship. Let’s say cruise ship from now on.)

      If the cruise ship is moving at a constant speed, it’s just as hard and just as fast to run to the back of it as it is to run to the front of it. This is the case even if the ship is moving at exactly your running speed, such that when you are running towards the back, an outside viewer would see you as being stationary to the sea (i.e. exactly like a treadmill).

      From the point of view of your muscles, you have no way of knowing if the ship is completely still, moving it the direction you’re running, or moving in the oposite direction.

    6. A dyno has resistance.  The cars tires spin the dyno, the dyno does not move on it’s own.   The dyno measures the power and torque applied by the wheels of the car.

      Your point about the treadmill not being the same as running on solid ground is valid though.

      1. Yeah, like Adam said.  When you take your car in for a smog test in California, they drive it up on to a dynamometer.  First thing they do is measure the vehicle’s weight at the drive axle, which allows them to calculate the amount of resistance the dyno will have to present the drive wheels in order to accurately mimic the torque load the drive wheels experience during actual on-the-road operation.  Those spinning cylinders on the dyno aren’t just free-wheeling; the resistance they generate is adjustable for this purpose.  The point of which is to measure the vehicle’s emissions output while “driving” at 15 mph and 25 mph.  Putting the engine under this load makes the test more stringent that the old 2-speed idle test, which was done in neutral with no load on the engine.

        All-Wheel-Drive vehicles can’t really use a single-axle dyno without zipping off and smashing through the back wall of the shop, so they still get to use the two-speed idle test, last I checked.

        1.  Ah. I was confused until I got to your second paragraph. I’ve only done one smog test in CA (I have New York plates again now) but they definitely didn’t put my car on a dynamometer – I drive a Subaru, which are all AWD.

        2. I’m having a hard time with some of this. I can only go by my own experiences. 

          Experience: Riding a mountain bike with front shocks. Never again. The front shocks absorbed to much of my forward momentum, I hated it, It was like sinking in sand. 

          The dynamo has been challenged for years. It’s antiquated. I don’t remember the magazine, but the article focused on more accurate measurements. Why? Listed  MPG’s are never accurate and are overstated. They are overstated because the condition is optimal,  the car’s body isn’t moving. 

    7. Physics (Newton basically) says that the only thing that matters is the speed you are moving relative to the surface. The speed the surface is moving relative earth, the sun or the stars is totally irrelevant.

      Next time you go by train (for a train is not that different from a treadmill), try running in the corridor both when the train is standing still and when it is moving at a constant speed. See if you can feel a difference.

      The real difference is instead in the hardness of the surface and the monotonicity of always running at exactly the same speed, with the same incline and without any turns. Compare that with trail running and I’m sure you see the difference that really matters.

      /Ms.C. Engineering Physics and a running coach

      PS When you start running close to the speed of light, then we can begin to talk about differences but only Ethiopians run that fast. Are you Ethiopian?

      1. The train is different, I end up somewhere else. But I assume you mean I would get on the train, walk a certain speed as it’s deck passes underneath me, and finally  exit in the same spot I got on. 

        But this ship train comparison isn’t working for me, they both move, I could walk on a treadmill on a train or a ship too. 

        I did think of one action that fits, swimming upstream. Is a treadmill the same thing?

        1. Absolutely. Swimming upstream is a perfect example: it’s just as hard to stay in place with a 5MPH current as it is to swim 5MPH in a lake.

          The stream is exactly the same as the treadmill.

    8.  Let’s take your rock climbing example.

      The wall is moving downward. You’re clinging to the hand and foot holds. As a consequence of friction (and gravity) you are also moving downward at the same speed as the wall.

      Relative to the wall, you are not moving. But relative to everything else, you are. Let’s say at the bottom of the wall (where it loops around like a conveyor belt, perhaps) there is a pool of water. If you simply cling to the rock wall, you will move downward with the wall and end up in the water.

      To stop this, you have to climb. You have to exert energy against the rock wall  and gravity to move upward in relation to the wall, and remain stationary in relation to the water (assuming you move upward at the same rate the wall moves downward). Otherwise you end up in the drink.

      The popular way to demonstrate the flaw in your thinking to is to picture someone on top of a train or in an airplane. Commercial airplanes fly at very fast speeds, approaching the speed of sound. If you stand up inside of an airplane and jump straight upward, why don’t you fly backward into the rear of the plane? Simple – you are also traveling at near the speed of sound. Your body has been accelerated along with the plane itself, and your inertia is carrying you along with the vehicle. Since no forces are acting against that inertia, you do not slow.

      Or take the case of standing on the top of a moving train. If you jump straight up from the roof of a speeding train, you land almost exactly where you were, for exactly the same reasons. The one minor difference is that the roof of a train is not enclosed, and consequently you are suffering the (very small) effects of friction as the air of the atmosphere rushes past you (or more accurately as you rush past it, but in terms of physics those are the same thing).

      There is no difference, in terms of physics, between running on a treadmill and running on solid ground. The only reason it appears different is because of your frame of reference.

      Imagine those moving walkways at airports. If you are unfamiliar with them, they are very long conveyor belts sets into the floor that serve the purpose of speeding up foot travel for passengers moving through the airport terminals.

      When you walk on the conveyor belt, you notice two things. 1) In relation to the rest of the building, you are moving at a higher than normal speed and reach your destination sooner than you would otherwise. 2) In relation to the conveyor belt you are standing on, you are moving at a regular walking pace.

      Both are true, and the effects are cumulative.

      Now imagine that you are walking down the same conveyor belt once more – but now imagine that the entire airport terminal is also on it’s own conveyor belt, moving at the same speed as yours. In relation to the rest of the terminal, you are walking at an ordinary speed.  If you stop walking on your conveyor belt, it will appear as if you had stopped moving altogether in relation to the terminal.

      However, if you then went and examined your motion from a frame of reference OUTSIDE the airport entirely, you would notice that both you and the airport terminal are moving in relation to the earth and other objects such as trees or buildings.

  3. It only runs faster than me because I’m old, and not being chased by an enraged cuckold.  So there.

      1. You wish!

        Coming soon to a theater of operation near you:
        The version to suppress demonstrations and hunt down liberal insurgents, wherever they try to hide.
        Eight legged for intense psychologal effect.
        Now with surround sonic weapon panels.

  4. The lack of air resistance is the only thing that makes running on a treadmill different from running on some perfectly flat (non-moving) surface.  (See: Galilean invariance.)  Of course, you could always remedy this asymmetry by using a train in place of the treadmill. 

  5. This thing isn’t a cheetah, we all know what their tops speeds are, clocked at 70mph. According to wildabouttheworld.com, the Six-Lined Racerunner, a lizard, runs at 18MPH. 

    These inventors have high expectations. I think a dead cheetah can run faster than 18MPH.

  6. Somehow, it looks both terrifying and silly.  Maybe because it looks like it’s running backwards, what with its chicken legs.

    1. It is an optical illusion, especially at fast speeds. It’s feet, if they are thought about as shaped “normally”,  are pointing backwards. Plus, it seems its legs change lengths, normally our perception tries  to fill in our innate abilities to identify an animal. A bear, it’s front legs shorter, you run downhill to get away from it. 

      For this to be a true mechanical predator, the biggest mistake would be to differentiate back from front. Not being able to know which end to be afraid or which direction it’s facing, is a sphincter factor of 10. For both.  

  7. It is the strange kind of fear the thing induces that I find interesting. Around 12mph when it starts to really get going and I get this visceral reaction that something that looks like that should not be moving that fast in an anthropometric fashion. Very uncanny. 

    1. I believe you’re probably reacting to the same thing I did, which is the transition to a “staggered” gait.  The pairs of legs stop moving in sync and all four hit the ground at different times.

  8. I don’t have to run faster than the headless Cheetah robot.  I just have to run faster than the treadmill.

  9. Bonus points to the inventors to make Edward Scissorhands pit bull into a practical robot.  Give those appendages a good sharpening (after he gets off the treadmill) and he could aerate your lawn while out exercising . . . in between shredding the neighborhood kids.

  10. One thing that is really cool is how the left and right legs shift slightly out of phase as the thing gets up to speed. Real four-legged animals do this too. I think it smooths things out and lessens the shock loads on the limbs and body, but I’m only really guessing.

  11. Eventually we’ll just have to show enemies videos of these things and they’ll lose all the will to fight. 

    Imagine this thing cruising towards you at even half the goal speed of 70MPH – or even worse, a pack of these things. It’s terror incarnate – you wouldn’t even have to give it any sort of weapon.

  12. its a good start, but the pre bent legs are not contributing to the stride. The cheetah will have full extension of the legs during their stride and there will be significant push off the legs at the extend. The spine also pumps up and down adding thrust to the push off on the back legs. There is some of that here, but not nearly as much. 

    A greyhound dog has a similar stride. The spine coils up to plant the back feet between the shoulders, and as the legs push forward the spine unwinds and adds the rotation of the hips to the push off.

    1. I saw on some nature program that a cheetah’s spine would travel something like 15 mph all by itself.

  13. Why is the military building robotic cheetahs on the tax payer’s dollar during one of the worst economic climates the US has ever seen? 

    1. Research funding is an economic catalyst. 

      If you disagree, can you let us know when it’s okay for federal agencies to start issuing funding for science again?

      1. How ’bout the same cash for a renewable energy project?  Instead of a killer cybercheetah to chase down the Middle Eastern soldier we’re fighting b/c we didn’t fund the renewable energy project.

        1.  pretty much. At this point it’s research for a useless military tool that will be gathering dust and any findings will be locked in a top secret vault somewhere.

          1. You people seem to forget how awesome Velcro is and who paid for its development. 90% of the things you use every day owe their invention to military funding.

          2. Jem: agree with you about the importance of fundamental research, but the 90% figure is just pure BS. Don’t overstate your cause, detracts from your credibility.

    2. My tax dollars are no longer at work.  They’re actively plotting my overthrow and destruction, in as creepily science-fictiony a way as they can imagine.  First the choregraphed swarms of quadcopters, and now, Ravage.

      Really, why didn’t they bother putting a head on this thing?  It could be completely nonfunctional, with fixed plastic jaws, chrome-painted balsa-wood teeth, and red LED eyes, and you’d still double your terror quotient.  Not that it isn’t terrifying enough, but you know, they’re still arguably marginally my tax dollars, and I’d really prefer to be hunted down and brought to ground by a robotic beast with at least a vestigial head.

  14. Really? They can create this mechanical marvel, but no one can figure out how to tie a plastic twist-tie to the ceiling so the disembodied ghost arm of Steve Jobs doesn’t have to hold the wire??

  15. External power makes it much less interesting. I’d like to see how fast they could make it with onboard power, and how far it could run on a charge. The video shows 18 MPH, and for the record, the human 400m world record is about 21 MPH. That means a human could outrun this guy for about 40 seconds. Then, you’re robot meat.

    1. The final version will have nearly unlimited range.  It will metamorphize human flesh into fuel. 

      I kidd!

  16. If they say it runs 18 MPH does that mean they cranked the treadmill up to 19 MPH and  it crashed in some spectacular fashion? Because I’d like to see that – in slow-motion, if at all possible. Kind of a schadenpistonfreude kind of thing, I guess.

  17. Is it using onboard battery power? If it’s cheating and receiving energy from those cables then I’m allowed a jetpack in this race.

    1. Yes it’s receiving external power, but it’s rather more akin to you being attached to a food tube during the race. It still has to create all its forward propulsion itself, unlike you with a jetpack.

      It’s connected to external power because they don’t yet have batteries small enough to attach to it that wouldn’t run out in 5 minutes.

  18. I get the fact that designs like this are really built as a testing and design platforms, not for explicit function.  I mean if we needed a robot that could do 20mph on the same terrain as this, I’m pretty sure something with 4 wheels would win.

    Besides I don’t need to out run it.  What’s it gonna do when it catches me, bump me to death?

  19. It can run 18 mph in a circle the radius of an extension cord.

    That’s not screaming you hear. It’s laughter from one foot further away.

Comments are closed.