"Jamming" Robot Gripper Gets Official: Article Published in PNAS

Jamming Robot Gripper

Remember that compliant "jamming" end effector unveiled by Colin Angle (iRobot CEO) at TEDMED 2009?  Even then, it was demonstrated picking up medication bottles, keys, and water bottles (a hand-held version was also demonstrated).  Well, it just got a whole-lot more official with the publication of "Universal robotic gripper based on the jamming of granular material" in the Proceedings of the National Academy of Sciences (PNAS).  The cool thing about this method of grasping is its relative simplicity: a rubber sack (balloon) filled with coffee grounds is pressed onto an object, it conforms to the object's natural contours, and the air is pumped out (a volume change less than 0.5%) to form a stable grasp-- no complex grasp planning required.  Be sure to check out the new video and photos!

Update on Oct 28th, 2010: John Amend, co-author of the PNAS article, wrote in to share this updated video with voice-over.


The latest effort is a joint effort between U. Chicago, Cornell, and iRobot.   Here is a video of the robot grasping all sorts of objects (eggs, LEDs, tubing, light bulbs, pens, etc.), pouring water, and writing with a pen.


Here is a brief diagram explaining how this method works, directly from the PNAS article.

Jamming Robot Gripper

ScienceMag discusses some of the limitations:

The hand works best on hard, dry, geometrically complex objects like screwdrivers and toy jacks. It has more trouble with flat objects like plastic discs and porous objects like cotton balls, because the air holes weaken the suction. It also can't grip anything bigger than half its size—the biggest items the team picked up were two one-gallon jugs of water. But the hand's true strength, according to engineering student and co-author John Amend of Cornell University, is its versatility. Aside from the limitations noted above, he says, as long as the gripper can fold about one-fourth of the object's surface, it can pick up just about any shape thrown in its path.

Jamming Robot Gripper

You'll notice that this latest version is very similar to the version Hizook covered previously (from Colin Angle's TEDMED 2009 talk):

IRobot Jamming End Effector
A close-up of the jamming end effector.  Right now it is (presumably) positively pressurized and in a liquid-like state.
IRobot Jamming End Effector
The end effector is (compliantly) pressed upon an object (medication bottle).
IRobot Jamming End Effector
Negatively pressurizing changes the end effector to a solid-like state, latching onto the object of interest for grasping / pickup.
IRobot Jamming End Effector
Jamming end effector lifting a plastic bottle.
IRobot Jamming End Effector
And a set of keys.
IRobot Jamming End Effector
Lifting a (dummy) patient's hand.
IRobot Jamming End Effector
Colin Angle demonstrating a hand-held jamming end effector to pick up a medication bottle.




Holy cow!  I just saw a comment on io9.com by Courteous_Gentleman showing a picture of Vac-Man. 

Vac-Man toy


I remember this toy (from the mid-1990s); my friend Kevin had one.  I also recall embedding objects in its hand using the vacuum pump.  It's funny how ideas work...

—Travis Deyle
It's an interesting solution to the problem, but from the looks of things it can't pick up an object from the side (hence holding the cup of water from the rim, which is awkward).  The glass was also half-full.  If the glass was full, any particulate matter on the end effector could potentially pollute the contents.  It seems like a good way to manipulate tiny objects, though.
Well it's properties suggests it's planned to be used in non-critical tasks. Like home robots or even toy robots. One application i can think of is a desktop robot, which organizes your, well desktop. It's not such a big deal if it can't pick up an object or can't drop it at the exact place you want and it's very easy to code. Also it's end effector is much safer than regular robot arms.
@yaang : this wouldn't do on a desktop, unless you neve use paper !

@ Robotbling, yaang, & feydaykyn:

Those are all good points.  The device certainly is not a panacea.  It's a (rather novel) specialized end-effector that is probably only applicable to certain scenarios -- much like the Dustpan Robot we built in our lab. As suggested by the limitations in the article, the gripper would be unable to grasp objects such as paper, credit cards, etc.  However, a very low-cost design that can pick up (hypothetically) 50% of nearby objects could still be very useful in many applications. 

Oh, it appears Popular Science's blog post has a picture of the coffee grounds inside the rubber sack / balloon (see below).  Incidentally, I disagree with their statement that its the "best gripper yet" -- that remains to be seen, and the tests in the PNAS article are insufficient to make such a bold claim (bah, sensationalism!).


Jamming Robot Gripper Coffee Insides



—Travis Deyle

We did this stuff back in the '70s:

Simpson, D.C. (1971) Gripping Surfaces for artificial hands. The Hand 3, 12-14

Kenworthy, G (1974) The Design and Development of an Artificial Hand Incorporating Function and Cosmesis.  Ph D Thesis, University of Edinburgh.

Good to see it's still going !!



I don't suppose you have PDF versions of those documents?  I'm unable to locate online copies, and haven't the motivation to dig through print library archives...

—Travis Deyle
Sorry - they didn't do PDFs in those days!  We didn't even have word processors!


That's tough to imagine, as it was about a decade before my birth -- but I believe you.  It's funny how that works; there are now folks alive who never lived without internet access... ;-)

It seems that SAGE has a PDF online of the first document (link), but that its behind a paywall not accessible online through my university account.  I thought that, perhaps, as an author you would have rights to post a copy...?  I'll just add it to my list for things to look up next time I hit the physical library.

—Travis Deyle

I saw on HackedGadgets that Carlos (of Carlitos' Contraptions) has created a DIY "Universal Jamming Gripper" using commonly available materials. 


Here is what Carlos has to say:

Since this is a rather simple principle, I thought I could quickly do one for myself.


  • A party balloon (silver in my case)
  • A piece of plastic tubing (I used a hose fitting I had lying around)
  • A piece of cloth
  • Ground coffee (I use some very old coffee I found in my freezer)
  • A rubber band.
  • Functional lungs (unfortunately, my vacuum pump is dead)


How to do it

  1. Fill the balloon with coffee (you might need to put it in a  vacuum chamber in order to be able to expand the balloon and thus stuff more coffee in it).
  2. Cover one end of the plastic tubing with the piece of cloth and hold it together with the rubber band. This will be used an an air-permeable filter that will keep the coffee inside the balloon and allow air to circulate.
  3. Insert the plastic tube in the balloon (so the filter is inside the balloon).


Now, you can connect the contraption to a vacuum pump in order to create a vacuum. you can also use your mouth and lungs in order to suck the air out. I had to resort to the latter method since my vacuum pump is broken :’(.  I am trying to source a cheap vacuum pump in order to improve the project and make it into a real robotic gripper. Any suggestions are welcome!


—Travis Deyle

Might I also suggest sonic vibration as another means of "liquifying" the interlocking particles.

A combination of several coincident approaches can result in very complex behavior for the particulate material.

Might I suggest that a linear actuator may also be possible, using this complex behavior , for a ferro-magnetic fluid within a magnetic field modulated by sonic vibration; or an analogous approach using modulated vibration within an electric field...

I am sure that you, and your graduate students, will have much to do in the months and years ahead.

—Daniel Gray

Of course, the air in the rubber membrane could be replaced by a low viscosity fluid.

Since air and other gases are compressible and liquids are relatively incompressible, this would be a choice for a gripper used underwater, surrounded by varying presures at different depths.

Space offers a different challenge.

If a low viscosity fluid could be found, stable over wide ranges of temperature and pressures, this might be a suitable choice.

The phase transition in pseudo fluids has been studied previously, but apparently with no similar applications.

It's good to see a fundamental design breakthrough like this -- and its useful applications -- in development.

—Daniel Gray

it seems quite possible to grip the glass from side too. it is a matter of providing the right suopport from opposite side of the gripping. may be by simple rubber support till the 80% of the height of the glass. then this baloon will expand and still can grip it.

if the creator of this video can do this exercise and upload the video it is welcome and will clear our doubts of this possibility too. waiting to see the video of side gripping the totally filled glass.

thanks for your doubt anyway.   ---Ajit-Pune India