Tattoos By Robotic Arm With Pinpoint Accuracy

Tattoos are an ancient art, and as with most art, is usually the domain of human expertise. The delicate touch required takes years to master, but with the capacity for perfect accuracy and precision movements, enlisting a robotic arm and some clever software to tattoo a willing canvas is one step closer thanks to the efforts of [Pierre Emm] and [Johan da Silveira].

They began by using a 3D printer modified to ‘print’ with a tattoo needle. Catching the interest of the Applied Research Lab at Autodesk, the next logical step was to use an industrial robot arm get a human under the tattooing machine — dubbed Tatoué — after scanning the limb in question and loading it into Dynamo, their parametric design environment to map the design onto the virtual limb.

[Pierre] put his leg on the line with the help of the handyman’s secret weapon to demonstrate Tatoué’s potential for some heavy metal ink. If you want a less permanent alternative to some robot art, a 3D printer can still sketch some impressive pieces.

[via Itay Ramot and Gizmodo]

66 thoughts on “Tattoos By Robotic Arm With Pinpoint Accuracy

  1. Reminds me of the 3d milling a design on a mouse hack using bump mapping and a probe extrapolated to a needle instead of a engraving bit and with a massive sponsor budget. Wondered when someone would join the dots up literally!
    I imagine you could do the same with the probe/bump map and a adapted mill/router/3d printer with a gun toolholder in it for 1/100th the budget…

    I’m rebelling by not having any tattoo’s personally.

    1. I’ll then make my fortune with laser tattoo removal vending machines. People can remove your tattoos the next day and have something else tattooed the next friday/saturday and then remove it again when they are sober. And such is the circle of life.

    2. Sooner or later we’ll figure out a way to tattoo e-reader “ink” that will let you turn individual dots on/off with a stylus or something. You could change tattoos on a whim, take notes on yourself, or disable them entirely for job interviews. The sky’s the limit.

          1. @Jii
            Haha, at first I thought I’d made a spelling mistake.

            Blink and marquee tags were depricated, as mentioned, but furthermore they’re no longer supported by any of the major browsers. There are browser addons for firefox–and possibly other browsers–that re-add support.

          2. RW and Blue Footed Booby:
            I know, the idea was not to make the text blink, but to joke about the idea of a blinking tattoo through the blink tags. I guess i failed twice.

          3. Fake it like [blink]that[/blink] it’s sort of semi-unofficially the default, nearly not quite ubiquitous, not completely uncommon way to denote html tags in fora that strip them.

        1. No edit :/
          It could work by the same principle of flip dots. Little magnetic bicolored dots kept still by electromagnetic charge inside a transparent medium. It should retain charge for a long time. The doubt i raise is that muscles rely and produce micro voltage, which can interfere with the display or vice versa.

    1. Really? I hope it isn’t. And i was horrified before!
      Hudson was far safer: ![knife game](

      CNC machines have only position feedback and planning… they assume that they’re going to go where they are expecting to at exactly the speed pre-planned… with potentially awesome force, and zero sensors to report on that force.

      I have heard second hand about that one time the proud owner of a new (refurbished) CNC machine hit ‘start’, on his first program, to discover that he hadn’t told it about the machine vice firmly bolted to the bed… After the machine calmly moved its spindle turret *straight through* the volume of said machine vice, without slowing at all, the vice was no longer there, but rather unevenly distributed about the room. (machine was a ~30T monster, with hydraulic servos, which was so old it had been converted from punch-card operation, and the guy operating it was a learner at G-code – a very bad combination.)

      Just to reiterate, CNC machines have absolutely *no* force feedback whatsoever, nor is G-code even able to encode the concept. They move to a plan, come hell or high water – up to the limit of their maximum force capacity.

      This is scary enough if you’re just using some nema 14 steppers via belts. It quickly becomes terrifying with much larger servo drives good for ~kW’s via ball screws.

      This article reminds me of that one on log splitting, with the wheel of doom, only far more dangerous. The technology to make these sorts of arms ‘automatically safe’ near humans doesn’t significantly exit, AFAIK. It would require sufficient low-latency, real-time, hi-framerate 3d scanner coverage for the thing to look where it’s going from all angles.

      Would be a great projection for a bunch of kinects. (or perhaps better yet a few pi compute modules with stereo cameras). AFAIK there’s nothing off the shelf that does it. Everyone with a serious machine just interlocks it so it can reasonably be assumed that any humans simply aren’t in the way. Real time movement planning from live measurements are an area of active research – just to avoid immobile obstructions.

      For this kind of project – essentially CNC surgery – one would be far better off with either minimum power actuators (like small steppers) or better yet, servo drives with controllable force movements. E.g. Tecnic ClearPath MC (PV or VC) in ‘torque command mode’.

      You’d want to be able to set things like ‘move in this direction, with maximum (x) N force, until you reach at most position (y)’. These can have asymmetrical force limits too (so the max allowed force is different if the motion ends up happening in a different direction).

      I believe *some* high end industrial robot arms do exactly this – usually so things can be assembled without breaking them (SCARA’s are one such).

      So I fervently hope this particular arm is one of the latter, and *less* like a CNC!

      It looks like they at least took the sensible precautions of limiting the robot’s maximum extent of movement to exclude most of the man’s body, and then providing force limitation by supporting the leg on (what looks like) an inflated wine cask bladder. (which is probably why the tattoo was flawed).

      I see people below are talking about improving the tattoo by holding the guy’s leg more firmly immobile. NO! EXACTLY WRONG! The only way to improve this about squishy living things is to have lower latency 3d scanning, preferably with some form of subsurface scanning to detect the appropriate depth to send the needle, just within the dermis.

      This is (one hell of) a stunt – but you’d have to be familiar with CNC’s and robot arms to fully appreciate how horrifying it is at first glance. Everyone who appreciated that was holding their breath, not to mention standing a safe distance away.

      The sad truth about most digital feedback control systems capable of force control is that they have a nasty tendency to be susceptible to a certain class of interference which results in an aliased signal causing an unexpected offset within the feedback loop. This usually results in a glitchy jerk when it happens, but that’s exactly the sort of unexpected movement you want to avoid near flesh. It also tends to corrupt the zero calibration, resulting in further movements happening where they shouldn’t.

      The vulnerability exists because of the common practise of using an antialiasing filter without sufficient rejection in the band starting just below half the sample rate, and extending all the way to the ADC chip’s natural input BW. For low latency digital control, so as to improve phase margin, it’s common practise to not have an antialiasing filter at all.

      The irony is that when such systems go haywire, it’s no fault of anything in the digital domain – just a consequence of ‘garbage in, garbage out’.

      If you look into the control systems actually deployed on CNC machines, you’ll find that where ‘digital’ feedback is used, it’s actually done via encoder or other ‘pulse type’ sensor, rather than via ADC. Or else the feedback is via analogue op-amp circuits which are immune to aliasing interference by virtue of operating in continuous time and having had their phase margin properly considered. (This latter is very common for hydraulically operated actuators, for example).

      1. I work extensively with both CNC machines and “industrial robot arms” and you are right, there is no force feedback on a stock CNC machine, but there isn’t on most robot arms either- not to the point where it would matter for this application anyway. But the device above is a Cartesian positioning system, not an ‘arm’ in any sense.

      2. “The irony is that when such systems go haywire, it’s no fault of anything in the digital domain – just a consequence of ‘garbage in, garbage out’.”

        – My professional life would be so much better if people understood this. Instead I hear “The robot just went crazy, we have no idea what happened”

      3. Lastly-
        “I believe *some* high end industrial robot arms do exactly this – usually so things can be assembled without breaking them (SCARA’s are one such).”

        Not all SCARA have this capability- SCARA is a HUGELY encompassing term. A power and force limiting robot, similar to a Universal Robotics UR5(UR10, or UR3 as well) or even *gasp* a Sawyer would have some force sensing abilities here, but even then I wouldn’t want one of those arms to tattoo me without some additional hardware.

    1. It’s traditional (my sister’s got one) for aspiring tattooists to practice on bits of pig skin from the butcher. She’s one of the minority in the family who isn’t vegetarian.

      She’s great at design, got a degree in graphic design, and did some weird stuff that would look good as tats for her degree show. but I think she’s lost interest in tattooing for a bit. Her fiance’s got a 3D printer too! Lucky me!

  2. It missed. The it lifted too far on the left side, that’s why the line fades out. Needs better Jig Or 3d scanning, not sure what part of the system dropped the ball. probably the duct tape.

    1. At least it didn’t go too deep. I’m guessing there’s no feedback for how much pressure is being applied the the subject so they probably just decided to err on the side of caution so dude didn’t end up with a nasty wound.

      1. aren’t tattoo guns like sewing machines where they have a “foot” thing to stop it from overpenetrating? i have no idea but on an industrial robot it seems like that would be easy to implement since it doesn’t need to “see” the leg.

        1. No, theres 3 layers to be concerned about here, the epidermis, dermis & subcutaneous layers.
          You want the ink in the dermis layer because it will bleed out when healing in the epidermis because its too shallow, and its possible to inject the ink too deep into the subcutaneous layer & you cause excessive pain and bleeding and run the risk of serious infection to the client also the ink tends to migrate because the layer has more blood flow. Google tattoo blow out to see this effect.
          The layer depths vary per person, setting up the gun and needle and first setting this up roughly then correcting this as a result of actually using the gun on a person is part of the skill of a tattooist. Which is one of the reasons why letting that scratcher in the bathroom at a party loose on your skin while drunk is a really bad idea…

        2. Yep there’s a screw to adjust how deep the needles go, on handheld machines at least. Also apparently tattoo-nerds get pissed off when you call it a “gun”.

          Big Clive, I think, dicks about with one in a video. Really he ought to spend more of his donation money on stuff to take apart, poundshop crap and Ebay tattoo machines are pretty predictable as to what’s inside and how they work.

  3. My Pict ancestors were renowned for covering themselves in tattoos,
    they also did a lot of other equally stupid stuff,
    then civilisation happened and the Scottish enlightenment followed.

  4. From the look of the heavy-duty taping they’ve done there, the weakness in this is obvious. I suppose the novelty value overcomes that, if you want a robot-drawn tattoo. Still, human flesh is a bit crap at staying in the same place, with the sort of accuracy a tattoo needs.

    Maybe they can 3D print something to hold the (user? victim?) canvas’s arm immobile, with a custom arm mould for everyone. Would be difficult to do anything with a bit of give behind it, like the body of most people who aren’t as taut and firm as they used to be.

    Right now, human tattooists have the advantage, they have visual input as well as output, and do a nice job, with stencils they can do some pretty accuracate stuff.

    Maybe in the future, somebody will sharpen up the pins on a dot matrix printer, or peel off the epidermis with a potato peeler, inkjet the tattoo on, then plop the skin back on again.

    Real men get Prince Alberts off 2-ton Unimation PUMA arms.

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.