5-Axis Printer Wants To Design Itself

The underside of the rotational base of the Gen5X 3D printer. A belt connects a pulley on the bottom of the stage to a stepper motor on the right side. The carriage for the stage looks organic in nature and is printed in bright orange PLA. The stage can rotate within the carriage which is mounted on two stainless steel rods connected to teal mounting points on either side of the printer (ends of the X-axis).

RepRap 3D printers were designed with the ultimate goal of self-replicating machines. The generatively-designed Gen5X printer by [Ric Real] brings the design step of that process closer to reality.

While 5-axis printing is old hat in CNC land, it remains relatively rare in the world of additive manufacturing. Starting with “a set of primitives… and geometric relationships,” [Real] ran the system through multiple generations to arrive at its current design. Since this is a generative design, future variants could look different depending on which parameters you have the computer optimize.

The Gen5X uses the 5 Axis Slicer from DotX for slicing files and runs a RepRap Duet board with Duex expansion. Since the generative algorithm uses parametric inputs, it should be possible to to have a Gen5X generated based on the vitamins you may have already. With how fast AI is evolving, perhaps soon this printer will be able to completely design itself? For now, you’ll have to download the files and try it yourself.

If you want to see some more printers with more than 3-axes, check out the RotBot or Open5X.

41 thoughts on “5-Axis Printer Wants To Design Itself

    1. Gregor looked at him sourly. “Notice anything different?”

      “Huh?” Arnold stared around the cabin, at Gregor, and at the Configurators. Then he noticed it.
      There were four Configurators in the cabin, where there had been only three.
      “You had them reproduce another?” Arnold said. “Nothing wrong with that. Just tell them to turn out a diamond apiece—”
      “You still don’t get it,” Gregor said sadly. “Watch.”
      He pressed the button on the nearest Configurator and said, “A diamond.”
      The Configurator began to quiver.
      “You and your damned pleasure principle,” Gregor said. “Repetition! These damned machines are sex mad.”
      The machine shook all over, and produced – Another Configurator.


  1. Now that there are more 4 and 5 axis printer designs, hopefully we will start seeing open source 5 axis slicers to complete with the proprietary DotX referenced. I understand Freddie Hong put together one in Rhino, but a stand alone program, even a basic one, would make this tech more accessible. Sounds like it would be fun to make one… if I had time… or money for a machine to test with… or skill :)

        1. It pretty much has to be that way round IMO – without the machines you can’t know the software actually works, you can simulate only so far before further refinement is likely to be wasted effort – either you missed something or the real world just can’t match the ideal one…

          Where without an ideal software stack you can still use the machines – add a nth axis so the print bed can be rotated to avoid printing overhangs or whatever and even if it requires a manual slice of all the separate elements and a little bit of gcode tweakery or writing the gcode entirely by hand to make a part it works. And of course it should still work normally with more developed software for the 3 axis style.

        2. I also believe it has to be hardware first and software second, since it needs to be known what capabilities can physically be executed. A person writing software with an Open5x-like printer in mind wouldn’t think about implementing an IDEXY collaborative tooling path feature, but would be a consideration for another person making a SecSavr Sublime-like printer.

        1. From what I’ve seen so far FreeCAD CAM is going to be what you need then. The CAD side of it I’m very comfortable with, and while its got some oddities so if you come from other GUI CAD platforms – so you will have to relearn how to do CAD a bit, it works well.

          The CAM side I’ve never actually used all the way, only simulated using – as I don’t actually have a machine yet myself to prove it works. But it seems to be perfectly functional. At least for 2.5D use, never tried simulating more complex stuff yet.

          1. It’s usable for 3d operations, but kinda quirky. (This may have improved since 0.19, which was the last time I tried 3d contouring.) It’s quite usable for 2.5D: I’ve made piles of parts with it. It handles putting dogbones in inside corners, tabs to hold material down, finishing cut offsets, cool-looking adaptive toolpaths, mostly automated helical plunges, about everything I’ve needed.

          2. After having worked with Solidworks, NX and even Fusion 360, I’ll simply re-iterate the “actually good” part and leave it at that

            With all due respect to the developers, FreeCAD (at this time) just ain’t it.

          3. @ThisGuy And when did you last use it – it is developing at rather impressive pace.
            The CAD side is no more perfect than the others, they all have quirks, but I’d say its no worse – just different. And the CAM side now exists.

    1. Actually, anyone who has build this printer, can have a free version of the 5 Axis Slicer! We only ask that you mention us in your posts, and that we are allowed to share movies and pictures of what you printed.

  2. We’re not there yet, but we are quickly approaching the viability of the “You wouldn’t illegally download a car?” argument the RIAA/MPAA used to throw out when they still were prosecuting middle school aged kids.

    The funny part was that the majority of us that were file sharing answered with a deafening “HELL YES!” when asked that question. Even as I rapidly approach middle age, I would still illegally download and print a reliable car if I knew I could get away with it. Especially considering how disappointed I’ve been with the last few vehicles I’ve had to live with. What is the point of mounting an oil filter in the hardest to reach place where it will be sure to make a mess and be impossible to grip? Why does the direct impingement engine have a timing belt instead of gears or chain? Why is the engine amazing and the drive train garbage?

    It will be interesting to see how organic/efficient designs become as software and 3D printing techniques improve. At some point we’re going to be able to feed parameters to a 3D production plant and its going to spit out what we need. Makes me wonder how many years away we are from Paradise Lost in Space, or the Butlerian Jihad.

        1. Unfortunately (for the dodgy doomsday peddlers), Grey Goo would need to outcompete the Green Goo that has had billions of years of head start in optimising for every niche it has ever been exposed to.

    1. Even when the “you wouldn’t download a car” ads were contemporary, every person I knew understood the false equivalence of the argument: we were all like “if I could try out a car without depriving anyone else of their car, just depriving a company of profit it wasn’t going to get anyway because I can’t afford their prices, I for sure absolutely would 100% download a car.”
      Was there _anyone_ who saw that ad and thought, oh, yeah, it’s just like stealing a car!

      1. These days I download stuff because I refuse to be extorted by shitty companies demanding I subscribe to their streaming service just so I can watch the one good show they’ve grabbed and held to ransom to try and bully people into subscribing.

        1. I do this too… But I also buy a physical copy where possible. Even if said copy stays in its shrink wrap in the back of a cupboard somewhere, it shows in some small way that there’s still a market for physical media.

          But I’m old enough to remember Metallica vs Napster. Never bought any of their stuff since, nor anything from Disney, since thy weaponised copyright law.

  3. What’s the point of the generative design? You know where your components are, connect them together. Optimized topology seems to same have the same result as non-optimized, all your loads have members going to them. Except now, it’s ugly!

  4. Optimally, you would end up with some sort of insane situation like the GOLEM Project from Brandeis University (http://www.demo.cs.brandeis.edu/golem/) where hardware and software develop together.

    Honestly, we really need some new tools in this space. I am not an AI expert, but it feels like one could partially automate the selection of fitness criteria in these generative systems in order to attain complex goals like “Take this model that requires 5-axis printing, generate g-code for it, then use that g-code to shape the requirements for a minimally viable printer. THEN take another 5-axis model, g-code it, then use the minimally-viable output printer #1 as a starting point to produce a printer that can output both models. Repeat as needed for edge cases.”

    The GOLEM Project slobbered on compute cycles to the point where developing the VM-in-a-screen-saver was a big part of the project so that the simulating environment could get computed at all. A penny buys you a LOT more math these days. Making a mutually-evolving system with more than one axis of fitness feels like it would require a nonlinear increase computing complexity, but we have the computing and we have better tools to use it.

    This project is cool, period. I just think that someone should call in Hod Lipson and Jordan Pollack to brainstorm with Ric Real.

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.