It is a shame that inkjet UV cure 3D printers haven't caught on with hobbyists. Inkjet 3D printers are awesome, the resolution is hard to beat, you can make parts with multiple materials as was shown in the article. Some can even print in color. But currently these machines tend to be very expensive, as in $250,000 and up. Worse, allegedly if you want to make full use of the printer's capability you have to sign over your IP to the company that made the printer.
Some people at MIT figured out how to build one cheaply and IIRC they even released plans for it online.[0]
You can also buy phone case printers on AliExpress which use the same inkjet UV cure process for about $1500[1]. They allegedly have some limited capability to do 3D too.
So it is possible to build 3d printers like this cheap enough that hobbyists could afford them. I suspect the reason they haven't caught on is due to patents, one may not even be able to sell a kit to make such printers
They haven't caught on outside of R&D and industrial uses because they're maintenance nightmares.
Think of an inkjet printer that prints honey, and the honey hardens when exposed to light.
The maintenance, the bad chemicals your exposed to in the printing, curing, cleaning processes, etc.
They're nightmares, you would not want it in your home. Same reason I don't have a sla printer, not worth it.
The projet 3500 is the coolest printer in this space if you do want one. It's cool be a use the support material is paraffin wax, so to clean and cure the parts you just melt off the wax in a low temp oven and clean the parts in an ultrasonic cleaner with cooking oil in it.
I tried to see what it take to build one.
My first issue is that there is very little information about how the tech work.
This article inkjet tech isn't the same one than the powder + inkjet: https://en.wikipedia.org/wiki/Powder_bed_and_inkjet_head_3D_...
> the resolution is hard to beat
Are you sure it's on the same level than the regular hobbyist resin printers ? Thoses have 35 µm to 50 µm resolution.
I am not referring to powder bed inkjet 3d printing, I am referring to making parts by inkjetting UV curable polymer. The 3D printer mentioned in the article, the cheap 3d printer previously made at MIT, and the phone case printer available on AliExpress all use this process.
>resolution
The inkjets used have similar DPI as those used in 2D printers, so resolution can be as small as 13-16 micrometers[0]
[1]https://www.rnd-tech.com/3d-printers-product-lines/polyjet/e...
> am referring to making parts by inkjetting UV curable polymer.
Yes, I said they were not the same, my point is there isn't even a Wikipedia article about it.
I don't see how technologically it can reach better accuracy than current resin printer that are based on lithography technics.
It's just a multi material resin printer but sounds way harder to do for an hobbyist.
I think this is one of the areas the whole "AI revolution" is lacking.
"Cognition", vision and generally the sensor parts are pretty advanced by now, ex: all the latest GPTs and vision Apis. But the actuator parts of the equation are pretty much the same as years ago.
I think a large part of this is the old industry saying: "Software is cheap/easy, Hardware is expensive/hard", These days "robots" pretty much means a robot arm that is either a clunky pile of parts with an Arduino for the hobbyist market or a multi-thousand dollar industrial arm to move card boxes.
I also don't know if human movement is a well understood problem yet, everything I've seen until now looks pretty much a better version of movement from a 's sci-fi robot movie ( I'm not discounting the difficulty of this, just pointing out )
Maybe this is the calm before the storm, now it's all "chat bots" and it's mostly inside our computers, when humanoid robots start to roam the streets and affecting the physical World it will be another ball game.
The sensors are not advanced at all. Things like pressure and texture sensing along with positional feedback are still extremely primitive compared to what human hands are capable of.
This might be a little progress towards miniaturizing muscles but we’ve still got a long way to go on the whole package.
I'm not sure the angels necessarily need physical robotic bodies, or just have to have consequences of their direct actions fed back, to get grounded. They can search the Web already, and kind of able to close a long loop when additional training is done upon them using the data contaminated with searches they make.
I agree. We'll have life-like androids for all kinds of entertainment (not AGI-controlled, just some fancy future-generation LLM) sooner than direct neural interfaces for full-immersion VR or stuff like that.
This reminds me of the real preemption to "West World". They had discovered cognitive machines and had them for some time, but printing "bodies" was the true break through that connected everything together. Pun intended.
Bones, ligaments and tendons... now they just need synthetic muscles! (I know there is research on that, but not sure if the current "state of the art" has performance comparable to "meat" muscles)
I think the technology to 3D print at the cellular-protein scale (microns to nanometers) is a ways off still. For anything like muscle, you'd need micromachines (comparable to actin-myosin) linking fibers (in muscle they're about 50um in diameter and a few centimeters long) that would work in coordination using some kind of energy (in animals the fuel is ATP) source.
Oh, so this isn't about optimizing torque/force distribution for axes? I always thought it'll be neat if I could move an arm parallel or orthogonal to an axis, simply by setting such predetermined set of servos to exact same values, thanks to such clever cross-coupling of joints. A bit like CoreXY, but requiring actual maths to set up, but computationally cheap to use.
When us humans animate, don't we tend to reveal such type of optimization that we can just add power to limbs to translate object, rather precisely, along cardinal directions as well as parallel directions, relative to our mental origin point? This is often explained as human brains being smart and doing fast IK calculations, but I doubt it.
But I ended out creeping out someone while trying to show how I can easily mouse-drag my wrist joint parallel to screen coordinates, so in cases above didn't apply to most, just disregard :p
I remember there was analogous project, that used inflatable muscles. It was at least 10 years ago, and at the time seemed like something from science fiction, but it was a real thing.
The status quo they describe is a printing process where material is deposited and then UV cured, then there’s some kind of scraping process. That seems like a hybrid of FDM and lithographic printing from a vat of goo a la FormLabs.
Industrial equipment uses this technique often for different purposes like glueing. Might seem obvious but the scraping is an artifact from the UV curing, it happens for SLA printers too. There’s a reason why the vat is a FEP film since it has lubricating properties like PTFE so each layer peels off easily. Also fun fact, that’s what carbon3D printers are so damn fast. They have a semi permeable film that lets oxygen through which creates a thin barrier of “cured yet viscous” resin between the film and part so they don’t have to wait to peel or scrape anything.
I wonder if there could be a way to use this technique to 3d print something like HASEL actuators for artificial muscles to go along with the tendons and bones.
I have been able to partially print them[0]. The electrodes we used could have been inkjet printed* and there is some potential of printing the dielectric fluid too. In the process I developed, the HASELs had to be filled with dielectric fluid during postprocessing, but it has been shown that it is possible to print closed volumes of fluid with inkjet 3d printers[1]. Although there is some risk the dielectric fluid could interfere with the curing of the polymer.
*We also were able to develop a technique for printing electrodes without modifying the printer, but this hasn't been published yet
Congratulations! If one were to try to commercialize this (assuming further development), do you think the Artimus Robotics patents would be a problem?
Hook it up to GPTs based on GPT-7, build the theme park... frankly, yeah, Westworld seems way, way more achievable in the short term than it seemed when I watched it way back when.
That said, while 3D printing like this may advance and be useful, I think more normal humanoid robots that are covered in a human layer is much more likely.
Some people at MIT figured out how to build one cheaply and IIRC they even released plans for it online.[0]
You can also buy phone case printers on AliExpress which use the same inkjet UV cure process for about $1500[1]. They allegedly have some limited capability to do 3D too.
So it is possible to build 3d printers like this cheap enough that hobbyists could afford them. I suspect the reason they haven't caught on is due to patents, one may not even be able to sell a kit to make such printers
[0]https://cdfg.mit.edu/publications/multifab-machine-vision-as... [1]https://www.aliexpress.us/item/2261799819029324.html