Ask Hackaday: Who Is Going To Build This Pneumatic Transmission Thing?


Disney research is doing what they do best, building really cool stuff for Disney and telling the rest of the world how cool they are. This time, it’s a very low friction fluid transmission device designed for animatronics.

From testing a few toy robotic arms, we can say without a doubt that servos and motors are not the way to go if you’re designing robots and animatronics that need lifelike motion. To fix this, a few researchers at Disney Pittsburgh have turned to pneumatics and hydraulics, where one joint is controlled by two sets of pistons. It’s extremely similar to the pneumatic LEGO, but more precise and much more lifelike.

The system uses a pair of cylinders on each joint of a robot. Disney is using a rolling diaphragm to seal the working fluid in its tubes and cylinders. This is an extremely low-friction device without any shakiness or jitters found with simple o-ring pneumatics and hydraulics.

The system is backdriveable, meaning one robotic arm can control another, and the other way around. Since we’re dealing with hydraulics, the cylinders (and robotic/animatronic devices) don’t need to be the same size; a small device could easily control a larger copy of itself, and vice versa.

The devices are fairly simple, with gears, toothed belts, and bits of plastic between them. The only unique part of these robots is the rolling diaphragm, and we have no idea where to source this. It looks like it would be great for some robotics or an Iron Man-esque mech suit, but being able to source the components will be a challenge.

You can check out the videos of these devices below, and if you have any idea on how to build your own, leave a note in the comments.


59 thoughts on “Ask Hackaday: Who Is Going To Build This Pneumatic Transmission Thing?

  1. I realize this is a little bit of a .. well, a hack, but seems like a condom would make a good rolling diaphragm at least for proof of concept. Wear would be an issue, of course, but as long as the gap it’s filling isn’t too wide, it should hold a decent amount of pressure.

    1. “Wear would be an issue, of course, but as long as the gap it’s filling isn’t too wide, it should hold a decent amount of pressure.” << you know that sentence would seem perfectly normal IF you didn't mention the condom

    2. hurr hurr, condom, hurr durr. Oh wait, I’m not actually a teenager.

      Anyway, if you can get the gap small enough, sure. Otherwise you’re going to get a lot of rubbing, and it’s going to require a lot of lubrication to prevent it from tearing. For a proof of concept, it’s a very good idea though.

      1. A syringe would work but there would be a much higher resistance to movement which would affect the amount and speed of transmission. I personally would use heat shrink rather than a condom however as condoms are rather elastic under pressure. using tubes of different sized for sort of pneumatic version of gearing could be interesting for feedback (large pair of reciprocating tubes that feeds into a bunch of little effectors).

    3. I don’t think so… I think they would be too stretchy. The industrial rolling diaphragms I worked with 8-10 years ago were actually kevlar-lined to ensure that they wouldn’t stretch so that the volume on the wet side of the the diaphragm stayed under control. (it was a sterile volumetric liquid pump)

  2. this may sound crazy but even though the results are a bit different the design is very similar to a patent that dana has on a transmission from back in the early ’90’s. I can’t find it right now because they have so many but it is similar.

  3. The use of the rolling diaphragm is very clever! I like this!
    Shame it’s proprietary, but hobbiests should be able to reproduce it non-commecially with basic home-casting materials.

    1. Is it patented? I didn’t do a search, but if it’s not patented then it’s fair game.

      I suspect based on no evidence that even though the application is unique, the principle of a rolling diaphragm pneumatic cylinder has prior art.

      1. Many designs of hydraulic brakes out there use, or have used rolling-diaphragms.
        Used to be common in motorcycle brakes. But not used much anymore because precision machining, and hard-wearing metal alloys have reduced the need for it (simplifying mass-production too).

  4. A way to create this could be similar to a hydraulic braking system in a car, mountain bike or motorbike, the idea being a hydraulic piston driving another hydraulic piston, one lever/pedal activated and the caliper piston being driven by the displaced fluid. If you attached two brake calipers end-to-end then it could theoretically do the same thing by pushing one piston to actuate the other.

    1. Also the return of the actuating piston will create a negative pressure differential which will then return the driven piston.

      I understand that a car or motorbike will make use of a master cylinder to do the driving but in its simplest iteration, the two calipers or any hydraulic actuator with a length of hydraulic line linking them will do the job.

      1. You could, but that’s not the unique part. “Disney is using a rolling diaphragm to seal the working fluid in its tubes and cylinders. This is an extremely low-friction device without any shakiness or jitters found with simple o-ring pneumatics and hydraulics.”

    1. The downside is the limitation of distance from the actor. The advantage that electrical impulses have is that they can be encoded, then transmitted over long distances with no loss of fidelity.

      This is probably a superior solution for the specific application that Disney is designing it for. Perhaps it would have a place in robotic surgery, or some other dangerous environment where high resolution, control and feedback are required, but I don’t think it’s going to kill electric robots.

      1. In some applications this can lead to great improvements. If your robotic application allows it, with this system you can offload the control gear off the moving gear. In other words, the servos/motors needed to control a robot arm do not need to be in the arm itself!

    1. Once any load (even friction) is on the system, air will pressurize in the lines, making the action delayed and limited in travel by taking up some of the displacement offered by the driver. By pressurizing the lines the air is already compressed above and beyond what the load can generate so it behaves as though the lines were filled with a fluid. Hose flex is a factor, but when using air, the air compressing has a much bigger effect.

      The genius is they can somewhat dampen the action by controlling the air pressure.

      1. Not only can they dampen the action by controlling the pressure, it wouldn’t be hard to make the dampening variable by adding a third cylinder to each line, with the position set via stepper motor to change the overall pressure in the line. You could vary system pressure in response to position of the input or output shaft, or based on the speed or acceleration of the input shaft.

    2. @Earl,

      They pressurize the air because air is a compressible gas. If they didn’t pressurize it, moving the source would simply squish the air in the cylinders and hoses without transferring much of the force to the load. By pressurizing the system, the gas has less ability to be compressed, and thus it has less “slop” in controlling the load. It doesn’t “tighten the hoses” as much as it “tightens the air inside the hoses.”

      Fluids, like water, hydraulic fluid, or brake fluid, are almost perfectly incompressible, so they transfer the force almost perfectly with no extra pressure. But because the water (or other fluid) is much heavier than air, moving them has to account for the inertia of the fluid, and has increased friction losses in the hoses, much more than a pressurized air system. To get a fluid based system moving, your power source (which starts from zero) has to keep adding power until it overcomes all the friction and inertia. That shows up as sudden starts and stops at the start and end of travel, which is bad for a robot you’re trying to make look lifelike. Fluids are also much more rigid than compressed air, and would show every slight vibration in the source power, including any steps in the stepper motor driving it. The compressed air is still a gas, and is still very compressible, and it is much springier than fluid. It might be more forgiving (and perhaps more lifelike) in an animatronic model.

      1. Adding a gas dampener to a fluid system will greatly reduce the sudden start-stop (for example: gas-charged shocks, or water-hammer in a big building, etc…).
        Also, adding spring-charged or flexible, shock absorbing connecting rods/arms/mounts better simulate the “stretch under force” encountered with living muscle-tendon-ligament systems.

  5. My first thought before seeing the comments was a condom as well. Might try finger cots, aka single digit gloves

    The great oracle Google shows us that rolling diaphragm are not uncommon or proprietary.

    looks to me like you could put a finger cot on a a few properly sized pieces of PVC and get something working. I doubt they would last for more then a bit of testing but for the cost it would be cheap entry into the field.

  6. At the bottom of the Disney Research page is a link to the conference paper:

    It contains the part numbers of the “off the shelf” diaphragm and diaphragm cylinder they used. The cylinder is a “Control Air 349-180-009”:

    I could make a comment about lazy article writers, but I won’t! :)

    1. Nothing. Absolutely nothing. Disney will or already has patented it and will sit on it. As a society, we have chosen to enlarge Mickey’s profits over advancing our entire civilization.

  7. I have succesfully used PVC pond liner a while ago to repair a boost pressure actuator on an old diesel engine. These also have a rolling diaphragm inside for minimal friction.

    The repair easily stood up to 30k kilometers on that car engine – so it may be a good choice for other applications as well.
    It’s just a matter of thermoforming the foil into a cup shape with a hot air gun…

  8. Rolling diaphragm…raspberry! What crap, reinventing the wheel. Fancy ‘o’ ring that won’t leak till it fails. It’s thick and it flexes in tight turns. Foreskin or turtle neck skin really.
    If servos are jerky it is bad programing and slop. Just look at Holeywood animation if you can see fluid motion. The people who have grown up on it don’t know motion…hint, it takes a few hundred frames per second and much more work. In watching the video I saw delay and un-damped oscillations that would flunk if used as a pitch bender on a synth.

  9. “Disney Research Pittsburgh” is Disney setting up Carnegie-Mellon students/labs with design projects, which this is. As previous comments have pointed out there isn’t much that’s revolutionary about this, but it is the usual press release “Revolutionary _______ will fix/lead to/cure _______ Real Soon Now”. Given our borked patent system, they’ll likely file on it but that doesn’t really mean anything.

    List of other stuff they’re working on:

  10. Low temperature differential sterling engines makers have found that a graphite piston in a glass cylinder makes for very low friction. I don’t know about how high a pressure it can hold however.

    1. probably fairly high, vacuum pump in mercedes for soft closure amd central locking is graphite internally. the air suspension is also rolling diaphram as per any car or truck with oe air susoensuin. mercedes, audi, landrover, and almost all?? “eurotrucks” im sure american trucks are still on leaf springs :p

  11. But why did they “re-invent the wheel” ?

    Pnumatic cylinders and diaphrams have been used as actuators for 150+ years.
    The “rolling-seal” isn’t new either, but the patents are from a long time ago so Disney doesn’t have to worry about that.

    I have a 100 year old camera remote-control that is almost exactly like the passive piston-to-piston actuator they’re demonstating… the major differences are that the camera remote is made out of ancient plastic and it multiplies force by having a larger volume stroke on the control cyler side.

    Oh well, I guess there isn’t anything new under-the-sun… everything is a re-embodiment or refinement of ideas that came before.

  12. A rolling diaphram is also known as a loud speaker spring. They’ve been used for over 70 years to mount the moving parts of a loud speaker. There can be no current patent on that idea. The weakness in the loudspeaker spring is that it expects a bit of stretchiness as the oscillating part approaches the limit of its travel, the inner diameter has to become the outer diameter or vice versa. rigitdity and flexibility are both required for best performance.

  13. I just bought an old ANIMATRONIC Moose (probably 10 year old technology) that used to be in a Bugaboo Creek Restaurant. It looks perfect but…. The Pneumatic cylinders are still inside it with the tubes leading out but that’s it. I need a way to “power” it up and control it… Plus need to figure out how to make it say inappropriate things. :-)
    Any suggestions?

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.