3D printed, pre-assembled robot hand


12 Responses to “3D printed, pre-assembled robot hand”

  1. LinkMan says:

    Please don’t let anybody who works at Cyberdyne Systems know about this.

  2. andygates says:

     Wow! Talk about maximising the “easily reproduce complex shapes” potential of 3dp!

    Now, I need one of those, some dacron line, a load of servos… and that Leap submillimetre Kinect that’s making the news right now. 

  3. Ultan says:

    It’s a clever design, though somewhat under-actuated. It has just 4 tendons per finger – 2 which pull the proximal phalange  in towards the palm and differentially allow moving the finger side-to-side, 1 which pulls the distal phalanges in, and one on the opposite side that pulls against the other three, allowing the finger to be pulled away from the palm. Real hands have more degrees of  freedom, but effectively this is about  optimum. There are a lot of complexities and tradeoffs in the design of anthropomorphic hands, very few have been produced that weren’t horrifically expensive and fragile prototypes. What with the limitations of 3D printing, this is still a fragile prototype, but much less expensive. (Still far from cheap, though, once you figure in the servos and controllers, etcetera. There’s a lot of etcetera.)

    I’d like to see a few more iterations on the 3D printer aimed at making an optimized design with parts that can be molded (which seems to be the design philosophy). I’d also like to see steel in the hinge pins and Teflon or HDPE in the bearing surfaces. Some sort of fixtures for attaching covers/pads would be nice – I’m thinking press-fit silicone. Conduits for sensor connections – using an existing bus standard such as I2C would make it easier to add more sensors later without adding more wires. I’d look at piezos first – simple pressure + vibration can tell a lot. Capacitive sensors are also easy and can be superior to human senses. Integrated controllers for the servos with force sensing could replace some of the other sensors at first. Some on-board processing so that it has a kinesthetic sense of its overall position/ configuration would be helpful, too. The electronics and software will end up being an even bigger project than the mechanical parts.

    • Christopher Chappell says:

      Hi Ultan, as the designer I have to agree with everything you said! My Mk1 hand actually did have steel hinge pins, and I’m thinking of going back to that for the Mk3, as they are superior to the printed ones.
      Unfortunately many design decisions have been made with trying to keep the costs low, and minimise the logistics required. However more interest from others in the project mean that I can achieve more whilst keeping costs low.

      I fully agree with your last sentence, and I’m hoping that by creating a standard set of mechanical parts, those more skilled in the electronics side than I can have something to collaborate on.

      • Halloween_Jack says:

        Even though LinkMan already kind of beat me to it, let me thank you on behalf of SkyNet.

      • msbpodcast says:

        The reason and my motivation for creating this prosthetic is a taxi driver I met years ago in Lawton Ok., who was missing both of his hands. He managed without them but it was a struggle as you can imagine.

        The design of the control electronics is relatively straightforward, if a bit non-intuitive.

        It consists of two sections of “homunculae” (as in Dr. Wilder Penfield’s neurological homunculae)

        They don’t have to be anywhere as complex as a real one. But you need to embed sensory feed back into the device, in some form of piezo electric, stained silicon devices, and the wiring for the feed back to, uh, feed-back, into a “skin” over/around the hand.

        In fact, of you’re hooking up a prosthetic device, its doesn’t even have to exist as you’re now looking to hook up controls and feed-back (possibly/probably wirelessly via bluetooth through the skull) to the device to/from a self-rewiring, evolving brain.

        The original implantation into/onto a disabled person’s brain would require some surgery and the wires would consist of a mesh of non-bio-reactive plastic and gold contact points. (That can’t be done by a maker project as I think most people would object to having the top of their skulls popped off with a hacksaw onto their kitchen table. [It'd be hard to explain the blood spatters to the police should anything go wrong.])

        The reason for embedding the mesh in the first place is to provide as much control/feedback to the brain as possible…

        • Ultan says:

           That’s … ambitious. While you’re working up to that, you might try TI’s ADS1298 chip to pick up muscular signals in the forearm (it’s also a good 8-channel EEG or ECG as well as EMG). Feedback through vibrations on the forearm is a good way to go, too.

      • Ultan says:

        Thanks for replying!
        The hinge pins at the base of the fingers are the weakest part – if you can add steel pins there for less than $10 or $20, I think  it might be worth it for the extra durability.

        Finger pads would be the other thing that would add the most utility for the effort required. Holes and/or grooves in the sides of the phalanges to hold the pads should be easy.  You could change the design and re-print, or try it out with a Dremel first.

        Here’s a rough idea for making silicone finger pads: use something like bakeable polymer clay to make a mold by pressing your fingers in, then pressing some appropriate tool such as the flat end of a drill bit into the side of the mold at the appropriate spot to make a molded protrusion in the cast finger pad that will lock into the hole in the side of the phalange. (You’ll probably end up using something other than actual fingers to get the  right width, but that’s the general idea.)Having 1-sided open molds printed should be pretty easy, too. (Fingerprint-like patterns are a good idea not only for grip, but also for later vibrational texture sensing. If using a clay mold, you might want to erase your own prints with some comb-like tool.)

        Silicone is the easiest material there is for casting, and it’s pretty cheap, too. I’d go with separate silicone pieces for each phalange to keep the fingers easy to bend. Pads are not only useful, they will likely also make the hand look more attractive and finished.

        [Edit: after reading the rest of the Anthromod blog, it's clear you know a lot more about molding than I do, and have enough equipment not to need to resort to crude hacks like the one above. Given the materials you have already worked with, silicone should be a piece of cake.]

        • Christopher Chappell says:

          Hi Ultan, if you write a comment on the blog I could continue correspondence. Speaking of finger pads, Shapeways has a new black rubbery substance, so I will try that.

  4. hakuin says:

    I see…..  and are We to conclude that this mechanical manipulator is the logical evolution of young geeks deriving the sensation of another’s palm by sitting on it until it goes numb?  Best to be somewhat wary of the more enthusiastic roboticists. A Fourth Law must be pending….

    • Christopher Chappell says:

      Well as long as it’s not used to abuse others I have no problem with what they use it for ;-)

Leave a Reply